The FOSS Photographer

Darktable vs Lightroom: Are we missing out?

2016-11-19T14:07:00.002-05:00

A few months ago, I posted a tutorial on photographing the Milky way, from planning to editing. In the tutorial, I showed the steps required for editing the photograph in Darktable. What I did not mention was that it took me almost 30–45 minutes to do all the editing in Darktable, trying to figure out the optimal adjustments to achieve the effect that I wanted. At the end of the process, I was quite pleased with the editing, though I felt that I should have done more while capturing the image. Maybe try an exposure with a wider lens with a larger aperture, get in more light in the image. Well, that would cost me quite some money, for a hobby that I only occasionally pursue. On a whim, I downloaded Lightroom for Android from the Play Store, copied over the camera DNG to my phone’s SD card, and started to play around with the image a bit.

Within a couple of minutes, just before going to bed, I had created, on my phone, an image that was more beautiful than the image I had spent close to three quarters of an hour on a computer. The results blew me away. You can see the difference between the Darktable and Lightroom versions below.

Lightroom

Darktable

To be fair in my comparison, I went back to Darktable, and used the adjustments that worked in Lightroom and made the exact same adjustments in Darktable, as best as I could. Even so, the image generated by Darktable is definitely desaturated (in the sky and the Milky Way, even with the saturation and vibrance sliders moved way up), and splotchy (on the ground and along the observatory wall), whereas Lightroom has made a lot of smoothing adjustments automatically under the hood to get me a pleasing image that is closer to what I wanted. Moreover, although the white balance numbers are identical in both software, the version is Lightroom shows colours other than blue.

Lightroom

Darktable

So, I ended up trying out the Adobe CC Photography suite, and I loved it. Now, I would not go so far as to say that Darktable is not a good raw processing software, in fact, it is great! However, I found that with Lightroom, I could save just a few more “bad” images, and this was important. When I say bad images, I mean images that were incorrectly exposed or blurry. Adobe has put in quite a lot of effort into great algorithms that produce pleasing pictures, although it is equally easy to go into territory where one’s images are a nausea-inducing vomit of rainbow colours.

I’m amazed by how good Darktable is, considering that it is a voluntary effort run by a handful of developers. The layout, the UI, the different modules, they are almost exact clones of Lightroom. Where Darktable differs is in applying multiple modules to specific parts of the image using masks. In Lightroom, one can paint over an area, then apply a number of adjustments, such as exposure, clarity, white-balance, etc. While I can technically achieve the same effect in Darktable using mask groups, Adobe’s approach is a lot more user friendly, and a lot more intuitive. Moreover, Lightroom allows me to preview the adjustments as I’m making them, unlike Darktable, where the drawing phase is separate from the adjustment phase.

So, are we missing out by using free and open source applications like Darktable over applications such as Adobe Lightroom? For the most part, I’d say no. I spent some time playing with some of my favourite images in Lightroom, and I could not tell which one was processed with Lightroom and which one was processed with Darktable when the images were exposed correctly in camera. On the other hand, for images that were incorrectly exposed, or could not be exposed properly due to some limitations of my camera and lens, Lightroom saved the day.

What does this mean for me? Well, I have bought a subscription for Adobe CC. Would I have bought the subscription if I was still in India? Probably not – the subscription is too expensive for an Indian income. However, in the US, it is the cost of one lunch for one month, and I’m happy with the product I get in return. At the same time, thanks to Microsoft’s Windows Subsystem for Linux (WSL), I am now spending most of my time programming on a Windows machine, so I don’t have to search for Linux-specific software any longer. As an aside, Darktable works wonderfully with WSL and XMing, in case you are a Windows user who wants to try Darktable.

Photographing the Milky Way

2016-08-06T10:10:00.001-04:00

In this post, I'm going to show how I created this image of the Milky Way, all the way from planning to processing.

Hartung-Boothroyd Observatory under the Milky Way on an August summer night

This post comes after a long break. The reason for the break is that my photography workflow has stagnated. Therefore, there really wasn't too much matter besides what had already been published on this blog. However, since the last time I posted a tutorial on Darktable, the software has advanced sufficiently – this warrants a new tutorial to show off some of the cooler features of the software – to become my primary workhorse. In fact, I really haven't used GIMP in the recent past.

Photographing the Milky Way requires planning. The Milky Way can be seen in the northern hemisphere only in the summer months. Therefore, August was really the last opportunity I had until March when the Milky Way would reappear. This is compounded by the difficult Ithaca weather, which presents rather gloomy overcast skies most of the time. The lucky break was just this past Thursday with a clear sky.

I envision a photograph of the Milky Way with an interesting foreground. I think it is the foreground that helps set the context for the Milky Way in the background, otherwise I'm just looking at a splash of colour against a black background with some pinpoints of light. Luckily, Hartung-Boothroyd Observatory is ideally located. It is an observatory, so it has been built on a relatively dark mountain and also makes a meta statement when photographing the galaxy. The fact that it is just a fifteen minute drive from my home is also very attractive.

To plan the shoot, I used Stellarium, a free and open source planetarium. It has an extensive database of cities and observatories, so I set the location to Hartung-Boothroyd Observatory, and the time to when I planned to be there. I was looking for the Milky way core being visible at the time, which it was – thank god – and that the moon would not be in the sky at the time. Now, it is possible to take a great photograph of the Milky Way with the moon behind the camera and low towards the horizon, that way it lights up the foreground just right and can be very interesting. However, having no moon at all makes it much easier. The information I gained from Stellarium helped me visualise the photograph in my head, so that I new exactly what I had to do once I reached the location. This helped me a great deal, because as you'll soon see, the Milky Way wasn't even visible to the naked eye.

I knew that the Milky Way would be due South, and the core of the galaxy would be down close to the horizon. This meant that I could not attempt a photograph with the arc of the Milky way above the observatory or something similar. I had to take a photograph with the Milky Way descending upon the observatory, or close to it. The next thing I had to decide was on how I planned to light up the observatory to make it stand out in the foreground. I recently bought a Metz 58-AF2 speedlight; I'd obviously use that, but in what colour? To decide this, I spent a few minutes taking photographs of a white wall in a dark room illuminated only using the flash and various colour gels. I knew that I would set the white balance close to tungsten in order to get a gorgeous blue sky, so I checked for each of the colours corrected for a tungsten white balance. I really liked pinks because they would appear a blue-magenta shade under the tungsten white balance setting; your creative opinions may vary.

So, the gear

Pentax K-5 with 18–55 mm kit lens
Tripod
Metz 58-AF2 speedlight
Flash gels
Torch
Remote shutter release cable (not used)
Heavy backpack to stabilise the tripod in case the wind was very strong (not used)
Some friends for company. It is really annoying to be up on a dark hill all alone, having people to keep an eye out for traffic and other contingencies is a good idea.

The rule for photographing stars is that you divide 500 by the focal length (adjusted to 35 mm format) to get the maximum shutter time. Photographing faster than this shutter time will prevent star trails. Don't ask me for the math, I'm sure that it should also include the sensor PPI or something, but I've found it to be a useful rule of thumb. Shooting at 18 mm on my Pentax K-5 gives me an equivalent focal length of around 27 mm, which gives me a shutter time of 18 s. This means that I could shoot either at 15 s or 20 s on my Pentax K-5. (A remote shutter release in bulb mode could have helped me get the sweet spot, but I'm an utter idiot and I forgot about this until after the fact.) To maximize the light entering the camera, I had my lens wide open at F/3.5 (the widest at 18 mm), and my ISO set to 1600 initially for a test shot.

Upon reaching the observatory, I was disappointed as I could not even see the Milky Way with my naked eyes. Turns out there are some factors that are not simulated by Stellarium. Always the optimist, I still set up my gear and pointed it in the general direction of where I expected to find the Milky Way. This is where the careful planning helped. The sky was just too dark to even see the stars through the viewfinder or in live view, so I resorted to dead reckoning, using my best estimates of what would be in the field of view for a 18 mm lens. I hoped to have the observatory to the camera right and the milky way dead centre. There was a tree as well to the left, I wanted that out of the frame. Once I had this approximately set up, I set my camera to manual focus, set to infinity, then took a test shot, ISO 1600, F/3.5, 15 s. This showed me that my framing was spot on, but the sky was still too dark. Therefore, I had to push up the ISO to 3200. I did not want to go all the way up to 6400 as I would have to deal with all the noise. While I'm okay with ISO 6400 for concerts and performances, the resulting noise looks a lot like salt-pepper (i.e. very white and black spots) that make it indistinguishable from stars in the night sky and much more difficult to clean up in post. My camera was also set to automatic long exposure noise reduction, which means that the camera takes a second frame immediately after the picture is taken, this frame is taken with the mirror down in order to get a black field. The idea is that the camera sensor develops characteristic noise which is not random when it gets hot, subtracting this black photograph from the original in camera is a great way to reduce noise caused by the sensor overheating. While this can also be done in post, I need to remember to take a photo with the lens cap on, which is much harder to do in the field.

Anyway, back to photographing. I set my ISO to 3200, F/3.5, 15 s. The shot looked good on the LCD, I could see the Milky Way faintly and I knew I could pop that out in post. The next step was to take photographs with the observatory (foreground) lit using my flash so that it would be more interesting that a silhouette. For this, I used my speedlight with a pink gel, positioned to the right and behind the camera. The speedlight has a guide number of 58 m when fully zoomed at ISO 100. Let me briefly explain the guide number and how that can be used to set the speedlight power.

Unlike the standard exposure triangle where the key metrics are ISO, aperture and shutter speed; a flash unit fires for so short a time as to make the shutter speed irrelevant. What is important is the flash power, as this is the light output of the flash. A guide number is simply the product of the F-number and the distance at which a subject is correctly illuminated by the flash firing at full power. This means that my flash unit firing at full power can illuminate a subject 58 m away when I'm shooting at ISO 100 and F/1. I was shooting at ISO 3200 and F/3.5, so the distance between the flash and the observatory should have been 58 * (3200 / 100) / 3.5 = 530 m away, which is clearly ridiculous. The way to fix this is to reduce the power of the flash. At 1/32 power, the distance is approximately 16 m, which is more like the distance I had in mind originally. The next part was a lucky fluke, but I forgot to account for the reduction in flash power caused by the gel. I used a Selens Follies Pink gel, which reduces the flash power by two stops, so my actual flash power should have been 1/8 and not 1/32. However, the lower power actually prevented the observatory from being too bright and distracting in the final image. In hindsight, I could have gone for slightly brighter flash, but these decisions are much harder when it is too dark to even see a single button on your camera. I handed my flash to a friend, asked him to aim it at the observatory and fire it on my mark. My friend was positioned so as to only light up the front face of the observatory and leave the side in silhouette, approximately 15–20 m away (more reasons for increasing flash power, with 20/20 hindsight). I set my camera on a 2 s self timer to prevent camera shake, and when the shutter opened, gave my friend a verbal cue to fire the flash at the observatory. Then review the image.

I just took four images, two each at 15 s and 20 s exposures, and with a Follies Pink and a Smokey Pink gel. I finally liked the version with the Follies Pink gel at 15 s, so that formed the basis of my final image.

We start off with the base image imported in Darktable. Darktable automatically makes some adjustments such as orientation, sharpen, and base curve upon import.

The first thing I want to do is set the white balance to tungsten to get the cool blue sky. That way, I can check all other adjustments while keeping an eye on the colour.

Then I push up the exposure by around 0.7 stops.

And add some contrast and saturation.

The next part is my favourite. Darktable recently added support for drawable masks, similar to the other software from the clay-company that shall remain nameless, but sounds a lot like the opposite of Darktable. I duplicated the exposure module, chose a large, soft brush and drew over the core of the Milky Way. I then pushed its exposure up by 0.7 stops and also deepened the blacks slightly

Next, I used the tone curve module to increase contrast even further in the galaxy core. Darktable saves masks, so I could re-use the same mask I used to correct the exposure on the core.

There's a lot of noise in the image, which may not be apparent in the screenshots, but it had to go away. I used the "denosie (profiled) tool, which is my go-to tool, but there was still a lot of noise. So I used the equalizer module set to the preset "sharpen and denoise".

Still not good enough, so go all the way, turning on "raw denoise", "hot pixels" and "chromatic aberrations"

Lens corrections!

Now that the image is coming together, I felt like the drawn mask for the galaxy core was just not wide enough. So I went back and fixed it by using a larger brush

Velvia is my favourite module to pop colours and get a look somewhat closer to film.

Some colour correction to make the highlights (the galaxy core) a bit more orange and the rest of the image a little more blue.

Second colour correction, just on the galaxy core (using the same mask as the exposure module), this time making the highlights just a bit green. I also used the mouse wheel to increase saturation, this has the added effect of increasing colour saturation on just the core.

Finally, something I forgot to do initially, adjust the shadows and highlights. This is, however, exactly the opposite of what I would normally do, and the defaults in Darktable. Typically, one would like to lift the shadows and make them brighter, or have the shadows positive, and bring the highlights down, or in the negative. However, in this case, I want the less bright stars to disappear and give me a cleaner image, so I push the shadows down to the negatives, and the highlights to the positives until I get an image that I like.

Remember when I said that I messed up accounting for the reduction in flash power caused by my gel? This is where I feel that the observatory is a bit too dark, so I need to pop the exposure on that as well. Create a new mask, and push the exposure up by 0.7 stops.

All in all, I pushed the exposure on the observatory up by 1.4 stops (0.7 + 0.7), and it is still a bit dark. My flash gel took up 2 stops of light. Guide number calculations work perfectly if you are smart enough to use them (unlike me).

That's it. I have no idea what the source of the spotlight at the bottom of the image is, but it sure looks cool right below the brightest part of the Milky Way.

This has been a longer post than most others on this blog, I hope you enjoyed it. Please do share your feedback in the comments below. Until next time…

My walk through Rivendell, or was it Watkins Glen?

2014-08-31T12:45:00.000-04:00

It's common knowledge that Ithaca is Gorges.

I really did not expect to be as blown away as I was with Watkins Glen. The gorge trail is breathtaking, and I felt like I was walking through Rivendell.

Watkins Glen, or is it Rivendell?

Watkins Glen

Images almost SOOC, minor edits in Darktable

Images on Flickr (1, 2), and on 500px (1, 2)

Taughannock Falls

2014-07-16T14:31:00.000-04:00

Within a few minutes (which I shall term amongst the more exciting ones in my life) of being published this photograph made it to the "popular" section on 500px.

Taken with just a polarizing filter to add clarity to the water and cut reflections, with minimal processing in Darktable.

New content coming up

2013-08-07T15:07:00.000-04:00

It's been ages since I've posted anything here, and that's frustrating.

I'm going to make it up to you folks who have been regularly following this blog. Stay tuned for some posts on awesome new tools that one can expect to see in GIMP 3.0 and Darktable 1.4.

Thanks for being such awesome supporters of FOSS

Concert Photography: Saxophone Sarod Tabla Trio

2013-05-18T16:02:00.006-04:00

Some time ago, the Cornell chapter of SPICMACAY (Society for promotion of Indian classical music and culture amongst youth) had a concert featuring George Brooks on the Saxophone, Salar Nader on Tabla, and Alam Khan on Sarod. Being a student with a DSLR, I took some photos at the concert, acting like a pro photographer, and ending up with no images to show that. :)

As always, here's the picture of the concert, with my workflow below:

With any concert, I need to fight low light, which means pushing the ISO up. A fast lens would help, but hey, I'm a grad student on a stipend. So, I open my kit lens as wide as it can, which generally means F5.6 at 55mm. A long focal length also helps, mainly because I need to take photographs without getting in the way of people listening to the concert. Shutter speed should be sufficiently low to prevent camera shake. This left me with the settings: ISO 1600, focal length 55mm, aperture F5.6, exposure 1/50 sec.

Okay, great, now that I've taken the picture, I need to process it some time, otherwise the folder is just a 4GB disk hog. So, load the image up in Darktable

and perform the standard corrections, like white balance

Exposure

Shadows and highlights

Equaliser

Then I move on to some basic corrections, like noise removal, lens correction, CA correction etc.

Then, I bump up the vibrance just a bit, with the Velvia module

And modify the tone curve to add some contrast

At this time, I felt that the image would look better in monochrome, because I felt that certain elements, like the background, and the big silver microphone were too distracting.

Which leads us to the final image, as processed by Darktable

So far, so good. The background still seems distracting, so let's get rid of it. Load the image in the GIMP,

I prefer to follow a non-destructive editing workflow, so I can always rectify any errors. This means not using the eraser tool, instead using layer masks to remove elements I don't want in the final image. So, the next steps involve masking out the background and the distracting mic in the foreground. Duplicate the image as a new layer, add a layer mask, filled with white (full opacity)

Next, add a layer filled with black and place it below the duplicated image layer.

The next step involves using the lasso select tool and selecting most of the background in the image, bucketfilling the layer mask with black (full transparency).

In the next step, I use my Wacom tablet along with a soft brush preset from GIMP Paint Studio (GPS) to clean up the rest of the background. Most of this just involved some careful painting.

Turn on the visibility of the black layer, to check the final results.

Just need a slight cleanup around the mic on the stage right (image left).

Finally, added a signature as a watermark.

Photographing Waterfalls, at Ithaca Falls

2012-11-21T21:38:00.001-05:00

I got some new gear, and I'm eager to share the results with you.

This is an image of the Ithaca falls, taken using an ND400 filter, a circular polariser, and an intervalometer.

This image was taken on a day with moderate clouds. Now, remember the sunny 16 rule, it's most handy. Since this image was taken in a gorge, in overcast conditions, I would require F4.0 and 1/100s at 100ISO. Now, I used a neutral density filter, which acts like a welding glass, reducing the amount of light through the lens. I used an ND400 filter, which reduces the amount of light to 1/400 of the original intensity. Then, to get a large depth-of-field, I reduced my aperture to F11. Now, F4->F11 means approximately 121/16 = 8 times less light, and my ND filter reduces the light 400 times. This means that I would need to take an exposure lasting 400x8x1/100s = 30 seconds. Of course, this exposure is well within the capabilities of my camera. However, I wanted to capture an HDR image, and hence wanted to push the exposure up to two stops more, which is 2 minutes. Hence, I used a cable intervalometer I purchased from Amazon.

To keep the camera steady, I used a rather cheap tripod by Polaroid. I'm on a limited budget right now, and I plan to upgrade equipment gradually over the next few years. It's foolish to spend around $1000 on equipment all at once.

At the exposure settings I mentioned (F11, 30s, 100ISO), the sky was extremely washed out, and completely overcast. However, I had taken a photograph minutes ago, at 4s exposure, and that served as the perfect image of the sky, because it was relatively free of clouds. The solution? A composite!

Let me describe my workflow in detail. Now, the part of the post written before this line was written just after taking the image, and nearly two months have elapsed before I have a little bit of time to continue with this post. So, just to keep things fresh in my head, I went through my workflow again, and it took around 30 minutes. That's quite low for one image, so you can guess that a minimal amount of manipulation was involved.

Let's start off by importing the folder in Darktable. You can use your favourite RAW processing software, and mine's Darktable.

I started off with the correctly exposed image which had details for most of the scene except the sky. Of course, the sky is an important element in the image, but it's not central. So, let's process the waterfalls first.

First step, adjust the exposure of the overall image. I like bright images, so I pushed the exposure up.

Next, I adjusted the white balance for a cloudy day. Notice that the water now has a weird yellow-green hue. We'll correct that in a moment.

Adjust the tone curve to add some contrast.

Next, I boost up the vibrance using the Velvia plugin.

However, I think it still needs some colour contrast. This picture was taken around the end of September (yes, wake me up!) and the trees were just getting the beautiful yellow fall colours. I want my photograph to reflect that.

Next, I increase the clarity using the equaliser plugin.

So far, so good. Now, to correct the highlights and shadows.

I've never really demonstrated the utility of the highlights and shadows plugin so far, and this is a good place to start. Of course, if you've had any history of image manipulation, you must have used something similar.

Basically, highlights and shadows help compress the tonal range of the highlights and shadows in the image; and this can be used to fit the highlights and shadows in the dynamic range of your display, print, and/or file. In this case, I found that the shadows on the rock face were too dark, so I pushed the shadows up. To show the effect of this plugin, I've split the display with the plugin turned off on the right of the split.

Now that the entire image looks brighter, time to get some details back into the waterfall. Now, that's the central part of the image, and I ought to do it some justice. So, I pushed the highlight recovery slider most of the way to the right, to recover the details in the water. Again, I'm using a split display to show the difference.

The sky still bugs me, so I go back to another image taken minutes ago, when the sky was relatively more clear. This was a shorter exposure, so the sky is darker, and if you look closely, you can see a bunch of guys jumping on the falls (Not sure if that is legal [:)]).

Open the image in Darktable

Perform the same basic image processing operations, like exposure, contrast, vibrance, equaliser;

and then switch back to lighttable mode.

Select the manipulated images, set export settings, and export the images. Then open them up as layers in GIMP.

I want just the sky from the image on top, so a layer mask is called for. Initialise a layer mask with black. Now, you can go ahead and mask the sky in with white, but painting in the entire sky is a tedious process. So, I use the Lasso select tool and select the sky region freehand using my mouse (yes, I need to buy a Wacom tablet, but, budget...)

Go ahead and bucket-fill the selection with white.

Of course, this result is not going to be perfect, so I used a Gaussian Blur on the layer mask

And then used a soft airbrush to blend in the tiny nooks and crevices that I missed with this bulldozer like approach.

So, what do you think of this workflow? How would you approach the same photograph? Any nagging questions? Feel free to leave a comment below.

Tutorial: Grunge up an image using Luminance HDR and the GIMP

2012-08-11T15:38:00.000-04:00

In this tutorial, I'll introduce you to my workflow which will allow you to "grunge" up an image using Luminance HDR and the GIMP. I'll show you how to make images look like this

from an image which looks like this without any processing.

Of course, I repeat, since the original image is just a DNG file, there's no way that I can say that this is how the image would have appeared straight out of the camera, as the camera would have applied some basic post processing algorithms to improve the image, had I saved it as a JPG. This is just the image with the "neutral" profile in Raw Therapee.

The original image lacks detail in the sky. The clouds in the sky were rather dramatic, but a linear output profile will not allow as much details as I would have liked. So, there's just one way to go: don't use a linear output, or even a gamma corrected output. Instead, I will tone-map the image to allow it to express all the dramatic details in the clouds in the sky.

So, the first step is to load the image in Raw Therapee with the "neutral" profile, add noise correction, lens correction, CA correction, crop, and then export the image to a 16 bit TIFF file. My camera outputs 14 bit DNG files, so converting to 16 bit TIFF should, in theory preserve all data. Note that TIFF is just a container, and the data stored can be compressed in both lossy and lossless fashion, as well as remain uncompressed.

The next step involves loading up the image in Luminance HDR. This is a single image HDR, contrary to the very concept of HDR. In fact, technically, it's not even HDR, but just a tone-mapped image.

since we are not making an HDR image, we don't really need any specific profile. Profile 1 will do.

Tone-map with Mantiuk '06 with contrast factor 0.1, saturation factor 2 and detail factor 1.0. As Mantiuk desaturates the image, saturation factor has to be pushed up to have an image with acceptable saturation.

Next, tone-map with Fattal, with alpha 1.0, beta 0.9, saturation 1.2 and noise reduction 0.2. Remember that Fattal introduces a lot of noise, so noise reduction is essential. However, the noise reduction is an iterative process, so very high values for noise reduction will lead to insanely high processing times.

This concludes the Luminance HDR part of the workflow. The next part involves merging the images to give us the desired image. For this, we'll fire up the GIMP, and load the base image (NOTE: This step is useful as Luminance HDR strips off the EXIF data from the images, so we need the base image to get it back in. However, as TIFF does not contain EXIF data, this step is unnecessary.)

Next, load the two images as layers, with Mantiuk positioned over Fattal.

Change the mode of the Mantiuk layer to multiply, and opacity to 80%. The Mantiuk layer contains more detail than the Fattal, but the Fattal layer contains more colour. Multiply helps merge the two. (Note: You can also try Value)

and overwrite the original image

This concludes the GIMP use in the workflow. Next, open the image in your favourite RAW processor (as long as it's FOSS [:)]), Raw Therapee in my case, because I was using Windows when writing this software. (Psst... Someone is working hard to port Darktable on Windows. Shh!)

I changed the profile to default-CC-BY-NC-SA, which is default with the addition that it inserts a CC-BY-NC-SA copyright notice in the image metadata.

Next, it's the usual changes, which include exposure, brightness, contrast, saturation, highlights, you know the deal...

A bit of vibrance never goes wrong

And that's it. Now, just export the image as a JPG

How did you like this tutorial. Share it with your friends and/or leave a comment below!

Unboxing my Pentax K-5

2012-07-31T03:15:00.000-04:00

Of course, you've seen pictures taken by my Pentax K-5. I just thought that I'd post the pictures of me unboxing the camera. The following pictures were taken using my Panasonic DMC-FS3.

If you stare long enough...

2012-07-27T02:07:00.001-04:00

If you stare long enough..., a photo by Skand Hurkat on Flickr.

Via Flickr:
You may see ghosts.

Another picture taken at the East Coast Park, and processed almost identically. Which one do you like more?

Sunset at the East Coast Park, Singapore

2012-07-19T11:21:00.002-04:00

I'm in Singapore.

Yesterday, I went to the East Coast Park, a beach on the east coast of Singapore, and took this picture

The image SOOC, would have looked like this:

Of course, since the image out of the camera was a DNG (digital negative (Raw, for those who understand the term)) file, there is really no way to say that an image would have looked like this. So, this is the image processed with no corrections applied.

Before I get to the processing, I'll briefly describe the process of taking the picture. "It's not the camera, it's the photographer" is the most famous quip amongst photographers, perhaps because it gives them importance. Let me tell you this, the camera matters every single bit. It's not just the megapixels, or the resolution of the sensor, or the dynamic range of the camera, it's also the amount of control that a camera offers. So far, I've been using a point-and-shoot camera for most of my pictures, with the DSLR occasionally borrowed from a friend. However, this image was made using a Pentax K-5, a real kickass DSLR I got my hands on recently.

This image was taken at the beach, a rather windy place with a lot of waves. When photographing waves, it's really easy to give a rather surreal appearance to the water by increasing the exposure time to insane amounts, like 30 sec, a minute, or even a few minutes. Such long exposures even everything out, and the water looks very calm, and yet maintains the colour of the waves, of splashing water, those ethereal white streaks.

So, I put my ISO down to 100, aperture to F9.0 to get a large amount of the image in focus (larger DOF) and also to reduce the light entering the sensor. Using the metering on my camera, that set the exposure time to 30 sec, good enough for this purpose. At such insanely long exposure times, a tripod is needed.

But I don't have a tripod. Before you start shaking your heads, let me tell you that I want a tripod. However, I'll be involved in a lot of travelling, and I don't want to lug a tripod everywhere. I will buy one in the near future, once my life is relatively settled :P

But if life at IIT has taught me anything, it's Jugaad, the art of fixing things. I propped my camera on my camera bag, set the self timer to 2 sec, and added a mirror lockup. This arrangement reduced vibrations, and gave a good-enough image. :)

So, I loaded the image in Darktable, and corrected the exposure. Since my camera's white-balance resulted in an image that was pleasing, I did not tinker with the white-balance. Otherwise, that would be the next thing that I would have adjusted.

Crop and rotate to correct the rotation (if I had a tripod, I could have used the composition adjustment in my K-5 to correct this in camera, before clicking the picture), and to crop out the bright lights in the extreme end (not having a tripod means that the composition is not exact).

Enable Velvia and boost the vibrance to 45%. Maybe a collection of good filters would have rendered this step unnecessary.

Next, I enable the equaliser and choose the sharpen and denoise preset.

Lastly, I add a frame and the watermark for the FOSS Photographer. I've saved this step as a style, so that I don't have to go through the multiple steps of setting frame colour, size, watermark type, position, etc.

So, what do you think of this attempt? Do let me know in the comments below!

Also, if you wish to share this image with your friends, you can do so on Flickr, Facebook, or Google+.

The Singapore Flyer

2012-07-15T13:00:00.001-04:00

The Singapore Flyer, a photo by Skand Hurkat on Flickr.

Via Flickr:
The Singapore Flyer as seen from Bayfront

I'm in Singapore!

I took this photograph when leaving Gardens by the Bay, and I must say, I really enjoyed the sunset while snapping pictures.

Damn the American system of placing months before days; it defies all logic, and ended up messing the date on this photograph, and all that I took over the two days before I noticed the error!

Marina Bay at Night

2012-07-09T05:44:00.001-04:00

Marina Bay at Night, a photo by Skand Hurkat on Flickr.

Via Flickr:
A panoramic view of Marina Bay at Night, overlooking Esplanade, the Singapore Flyer, the Marina Bay Sands Hotel, and Merlion Park. Merlion, unfortunately, was covered.

So, I'm in Singapore, and I thought that I'd leave you with this panoramic view of Marina Bay at Night. You may remember that I had used another set of images from the same spot, taken more than two years ago, in my tutorial on creating panoramas using Hugin. This is the same place, two years later.

Let me know what you think in the comments below.

GIMP Tutorial: Cross processed effect to change colours on an image

2012-06-26T03:50:00.000-04:00

In this post, I'll show how to make an image interesting. I'll show how I obtained this image:

from an image which looks like this:

The first step is to figure out what changes are required in the image. In my opinion, the blue sky is distracting, and I'd like to change the colour, so that the branch and the red leaves pop more.

Actually, I thought that a cross-processed theme would go well with this image, so here goes.

Cross processing is rather like the "retro" look, that I posted earlier. However, we would introduce some twists on the way.

Load the image in the GIMP

Colours -> Curves

Modify the curves as shown.

Duplicate the layer

And desaturate it based on Luminosity

Change the mode of the layer to Soft Light.

Create a new layer. Let's name it "Tint".

Fill the layer with a shade of yellow.

Change the layer blending mode to "Overlay" and opacity to around 75%.

This image is rather acceptable. But when I say cross-processing effect, I mean pushing the limits of technology, to create images that would have been when technology was shitty. So, I go ahead and add some colour leaks. Create a new layer group. I've named it "Vignette", but that's a misnomer. Sorry!

Create different coloured gradients. I use a non-destructive workflow, so I don't rely on the "undo" button. Hence, a new layer for each colour leak/gradient.

And that's it. Of course, you can add some random noise, or film-grain to improve the effect. Remember, you're limited just by your imagination. :)

HDR Tutorial: Living Room in Casa Loma, Toronto

2012-05-26T23:55:00.000-04:00

Here is another tutorial on making HDR images. Today's subject will be the living room in Casa Loma, Toronto. We will achieve this image

Casa Loma, Toronto

from this image

I will first describe the concept of HDR imaging and then go on to the process of producing the above image from the actual photographs. If you understand the concept, you may skip to the next section.

HDR Imaging

Consider the photograph I displayed above

This scene has a lot of elements making up for a good composition. However, the image fails to do justice to the scene. So we can try to reduce the shutter speed a little to get the following image

While this image provides a lot of detail on the room, it loses all the details on the candles and curtains, leaving the regions over-exposed. We can capture the details of the curtain and the candle by reducing the exposure (increasing the shutter speed), to get a third image.

Each image has a limited dynamic range of lighting it can capture. Hence, a single image fails to capture all the details of this scene. However, these 3 images together provide us much more information of the scene than a single image can capture. Using these three images, we can create an image much closer to what the eye sees, like the following image

Such an image is produced using special low dynamic range (LDR) projections of high dynamic range (HDR) images which are a combination of multiple "true" LDR photographs and are often called HDR images.

Software Used

Luminance HDR (formerly qtpfsgui), GIMP and DarkTable

Process

1. Capture a set of 3 or more images. Many cameras come with an auto-bracketing feature. If your camera has a +-2 EV auto-bracket, use it. If not, just adjust the exposure compensation to +2, 0, -2 and capture a photograph at each. For this scene, we have the following photographs.


+2 EV	0 EV	-2 EV

2. Now begins the processing. For producing different projections of an HDR image based on these 3 images, we will use Luminance HDR. Here, I have produced three different images:

Mantiuk '08

Fattal

Ashikhmin

You can see read more on how to produce these in a previous blog post.

Neither of these images is exactly what we want out of the result, but each of them provides an important attribute to the image:

a. Mantiuk '08 provides the right amount of lighting at the window and the candles.

b. Fattal provides detail across the image along with good color saturation. It can provide a lot of detail, but at the expense of grainy features. Hence, the noise reduction offers a trade-off between detail and noise.

c. Ashikhmin (with Eqn 2) gives good lighting across the image, making it a desirable base for layering in GIMP.

3. Layer the images in GIMP: We layer the three images in GIMP. Fattal layer uses the multiply layer over Mantiuk '08 and Ashikhmin layers which are used in normal mode. The multiply operator tends to increase the contrast of the image, which is desirable in this scene. The Ashikhmin layer provides a good base with low noise for the image unlike the high noise Fattal layer.

4. Some final touches in DarkTable (Refer to Tutorial) and we are done.

Note the detail in the curtains which are visible only in the low exposure (-2 EV) image and the details of most of the room, found in the high exposure (+2 EV) image.

If you found this tutorial helpful, leave a comment below and share it with others on Google+ and/or Facebook.

An introduction to Darktable

2012-05-17T04:12:00.000-04:00

I decided that I'll do something different today. I thought that I'd introduce a software that I had not discussed so far, but which is extremely relevant and useful.

Darktable is a raw processing software, but it can be used for a number of image manipulations. It is available only on the Linux/Mac/Solaris platforms, which may be a limitation, but the entire processing is high precision 4x32 bit floating point, GEGL based, with GPU accelerations possible via OpenCL. (Start drooling now!)

I have so far refrained from using Darktable because it's *nix only, and I generally use Windows due to poor hardware support on Linux (for my computer). I guess I cannot go on ignoring it any further. This tutorial will demonstrate a raw file processing workflow using Darktable, and ought to demonstrate just how easy is it to get those perfect effects using minimal clicks.

The workflow will show how I obtained this image

from a image which looks like this:

The first step is to fire up Darktable and import the folder which contains the camera raw files. Note that the import does not copy files, it simply imports files into Darktable.

The next step is to locate the image of interest using the thumbnails. At this time, I would like to explain the different modes in Darktable. The one shown above is the lighttable mode, which displays all images. The other mode is the darkroom, which is used to manipulate a particular image. Double click on the image of interest to enter the darkroom mode.

The first thing in any workflow is to correct exposure. But before you do that, it may be advisable to take a snapshot, which stores the image as currently displayed. This is useful for comparing results later on. For the exposure, you can select auto, and then the area of interest in the image, and Darktable will try to automatically adjust exposure

however, it's often off the mark. I prefer to manually adjust exposure settings.

I then crop out the image of interest, using golden means as guides

Which is followed by adjusting the tone curve

I then notch up the vibrance a bit

The next tool is rather interesting, unique and useful. The colour correction may seem daunting at first, but it's easy once you get the hang of it. Basically, it allows for different tints in the highlights and the shadows. You need to draw a rectangle (basically, provide two points) to indicate the tint in the highlights and in the shadows.

adjust the rectangle to adjust the tint. Note that the white bars indicate the tint in the highlights and the dark bars indicate the tint in the shadows and use the mouse wheel or the slider to decrease saturation, just a bit.

An interesting feature in every Darktable plugin is the ability to treat the output of the plugin as a layer and blend the layer using blending modes similar to those found in the GIMP. The opacity, and blending modes can be set. In this case, I set the blending mode to normal, and opacity to 70%.

Add a lens correction profile for the lens used.

Then head over to the equaliser tool.

you can choose from a number of presets like sharpen and denoise (selected). Or you could go ahead and manually tweak the equaliser.

The next step is to add a vignette.

To add a nice finish to the image, I add a frame

and a watermark

while the default watermark is good enough, I prefer my customised watermark which displays title and creator information from the image metadata.

I then add the soften filter, which works by merging an overexposed, blurred image with the current image

The default settings are a over the top, so I reduce the mix to a lower value

Just to show how far we've come...

Finally, hit 'Ctrl+E' to export the image with current export options, or switch to lighttable mode, and set the export options

And that's how you process an image with Darktable. Though this was not a raw image, the steps to process a raw image are much the same.

PS: Do let me know what you think of the whole Darktable vs Raw Therapee business. Which one would your prefer. Leave a comment below.

Understanding GIMP layer groups

2012-05-15T03:08:00.000-04:00

This post is a very short introduction to layer groups, one of the latest additions to the GIMP. Since version 2.8 came out, I guess that there has been a lot of confusion regarding layer groups, and what they can, and cannot do. While this tutorial cannot claim to demonstrate all the uses, I hope that it will provide remarkable insight into the additional processing power that layer groups offer.

I'll show how I obtained this image

from this image

Of course, this image was already processed much to my liking in Raw Therapee, so the following manipulations are more for academic rather than aesthetic reasons.

To begin with, I load the image in the GIMP.

I then add a new layer group and name it "Pop", for reasons that will become apparent soon.

I duplicate the original image twice,

and add the copies to the Pop layer group.

The next step involves desaturating the top layer in the Pop group

based on luminosity

I then change the mode of the desaturated layer to overlay

and blur it using Gaussian blur

IIR method with a size of 25x25 px (of course, this depends on the size of the original image)

As the final step, I change the mode of the Pop layer group to multiply

to obtain the final image

Of course, this tutorial is overly simplistic, but I hope that it provides enough insight into the power of layer groups. Without layer groups, I'd have to start with the same three copies, but after blurring the topmost desaturated layer, I'd have to merge it down, thereby introducing a destructive step in my workflow.

Layer groups get even more interesting in complex examples. Typically, I might want to change the colour of some region using a colour layer mode, then dodge/burn using an overlay/soft-light mode, and use the resulting image to merge with another image; the possibilities are endless; limited just by your skill and imagination.

Do share your experiences with layer groups. In case you have any tutorial that showcases the power of layer modes, do leave a comment below.

Choreography

2012-04-11T13:16:00.001-04:00

Choreography, a photo by Skand Hurkat on Flickr.

An image of a fountain by Place de la Riponne, Lausanne, Switzerland.

This is not a tutorial, mainly because the image is almost SOOC, with minimal exposure corrections in Raw Therapee

HDR Tutorial: Sunset over the Lausanne Skyline

2012-04-01T11:33:00.000-04:00

Another HDR tutorial. In this tutorial, I shall show how to create HDR images like this one

from the below three images

To make this HDR, fire up Luminance HDR. Select the option to create a new HDR image

In the wizard that pops up, click on next.

Load the three images in Luminance HDR. As these images were taken hand-held, and not using a tripod, they need to be aligned. Make sure that the auto-align option is enabled, and that the method is Hugin align_image_stack

Click next and let Hugin align images for you.

A window pops up, which shows the aligned images. Make sure that the images are aligned. Then, click next.

A window pops up where you can select your HDR creation profile. Profile 1 works very well, so I'll stick with that. Click next.

Now, save the HDR image that is displayed. If you wish to manipulate the image any further, you can start at this point.

Set the tone-mapping operator to Ashikhmin, simple, equation 2. Make sure that the "Update current LDR" option is de-selected.

Hit tonemap to get the tone-mapped image.

and save the tonemapped image immediately.

Set the operator to Fattal, default values work well. Hit tonemap to get the second image.

save it.

Now, we're done with Luminance HDR. Fire up the GIMP.

Remember that Luminance strips all EXIF data from the images. To get it back, load the 0EV image into GIMP.

then add the two images generated through the Ashikhmin and Fattal algorithms respectively as layers.

At this point, we don't need the base image (the SOOC 0EV), so we delete that layer.

Make sure that the Fattal layer is above the Ashikhmin layer, then change the layer blending mode of the Fattal layer to Multiply.

Export the image. At this time, the image is rather good, but let's try to improve it further.

Load the image in Raw Therapee.

I changed the profile to Insert Copyright, which basically is Neutral, but it also inserts my copyright information in the image.

Then select auto-exposure. Generally, it is good enough. In this case, however, I had to increase brightness to 5 points.

Modify the tone curve

Increase vibrance just a bit (5 points)

Now, I crop the image to get rid of the ugly borders that crept into the process with Luminance HDR (qtpfsgui), mainly because the images were not aligned. While aligning the images, Luminance HDR replaced the missing parts in each image with black, which lead to the defects at the border of the image. These defects must be manually cropped out.

Lastly, I add IPTC metadata, which includes headline, caption, tags, etc.

and then send the image to the queue for processing.

Head over to the queue tab. I keep the autostart feature off, as it really slows down processing if I have multiple images to process. It's a simple matter of starting the queue processing operation.

And that's the entire workflow I use to get an HDR image. Okay, so you Photomatix users will probably think that it's a lot of work to do, but tell you what, I like the freedom it gives me. Even so, pro HDR photographers like Trey Ratcliff or Victor Lacken anyway do a lot of post even on pictures out of Photomatix.

Lausanne is a really beautiful place to visit. The next time you are in the area, do make a point to visit Lausanne.

As always, feel free to leave comments below, or share this tutorial on your favourite social networking website.

Get the software mentioned in this tutorial from the links below.

GIMP script for the Lomo effect

2012-03-06T01:56:00.001-05:00

So, here's a GIMP script for achieving the Lomo effect. The script will enable you to create pictures that look like this:

from an image like this:

The original image was taken using the camera on my phone (LG P500), a really sad 3.2 MP camera. Nevertheless, I think the image makes an ideal example for demonstrating the Lomo effect, which is why I included it.

If you search on the internet, you'll probably find a number of tutorials on how to fake the Lomo effect. The trouble is that they don't seem to agree much with each other. Bummer! So, I came up with my own technique (if you feel that the technique is flawed, let me know, I'll correct it) based on what I thought gives the effect seen in many Lomo photographs.

Basically, I increase contrast on the red and green channels, decrease contrast on the blue channel; all using curves, then increase the saturation and add a vignette.

Without much ado, here is the script:
UPDATE: The code is now available on my GitHub repository.


;An example script for the lomo effect.
;This script increases contrast on the red and green channels, decreases contrast on the blue channel, and increases saturation on the image. It also adds a Vignette.

(define (script-fu-lomo-effect img drawable)

        ;Create a new layer
        (define curves-layer (car (gimp-layer-new-from-visible img img "Curves")))
        (gimp-image-insert-layer img curves-layer 0 0)

        ;Modify the curves
        (gimp-curves-spline curves-layer HISTOGRAM-RED 8 #(0 0 60 30 190 220 255 255))
 (gimp-curves-spline curves-layer HISTOGRAM-GREEN 8 #(0 0 60 30 190 220 255 255))
 (gimp-curves-spline curves-layer HISTOGRAM-BLUE 8 #(0 0 30 60 220 190 255 255))
        ;Increase the saturation
        (gimp-hue-saturation curves-layer ALL-HUES 0 0 50)

        ;Now, add vignette layer group
 (define vignette-group (car (gimp-layer-group-new img)))
 (gimp-item-set-name vignette-group "Vignette")
 (gimp-layer-set-mode vignette-group OVERLAY-MODE)
 (gimp-layer-set-opacity vignette-group 100)
 (gimp-image-insert-layer img vignette-group 0 0)

 (define img-height (car (gimp-image-height img)))
 (define img-width (car (gimp-image-width img)))
 
 ;Now, for the vignette layers
 (define vignette-white (car (gimp-layer-new img img-width img-height RGBA-IMAGE "White" 100 NORMAL-MODE)))
 (gimp-image-insert-layer img vignette-white vignette-group 0)
 (gimp-context-set-foreground '(255 255 255))
 (gimp-edit-blend vignette-white FG-TRANSPARENT-MODE NORMAL-MODE GRADIENT-RADIAL 100 0 REPEAT-NONE FALSE FALSE 1 0 TRUE (/ img-width 2) (/ img-height 2) (* img-width 1.5) (* img-height 1.5))
 (define vignette-black (car (gimp-layer-new img img-width img-height RGBA-IMAGE "Black" 100 NORMAL-MODE)))
 (gimp-image-insert-layer img vignette-black vignette-group 0)
 (gimp-context-set-foreground '(0 0 0))
 (gimp-edit-blend vignette-black FG-TRANSPARENT-MODE NORMAL-MODE GRADIENT-RADIAL 100 0 REPEAT-NONE TRUE FALSE 1 0 TRUE (/ img-width 2) (/ img-height 2) img-width img-height)

        ;Refresh the display
        (gimp-displays-flush)
)

(script-fu-register "script-fu-lomo-effect" "Lomo Effect" "Apply a Lomo effect to visible image" "Skand Hurkat" "Skand Hurkat" "2012" "RGB*" SF-IMAGE "Image" 0 SF-DRAWABLE "Image to apply retro effect" 0)

(script-fu-menu-register "script-fu-lomo-effect" "<Image>/Filters/Artistic")

The script is registered under Filters > Artistic by the name "Lomo effect".

As always, I look forward to receiving some feedback. Take some time to hit "+1" or share it with your Facebook friends or twitter contacts. Or leave a comment below.

Photographing Urban Decay - Take 2

2012-03-05T02:03:00.004-05:00

This post follows my earlier post on photographing urban decay. Basically, this process differs slightly from the "all Raw Therapee" based workflow of my previous post, and involves the GIMP.

As with all my posts, I present the finished goods...

The Raw Therapee workflow is very similar to the one outlined in my previous post; except that I don't desaturate colours, or change the RGB curves or white balance.

I load the processed image in the GIMP, duplicate the layer, desaturate it, set the opacity of the desaturated layer to around 45 percent.

Then I just fire up my "Retro Effect" script.

And that is how I create my second take on Urban Decay.

Photographing Urban Decay - Take 1

2012-03-05T01:55:00.000-05:00

Photographing is a tricky art. A photograph must convey to a large section of audience the exact thoughts that went through the mind of the photographer when (s)he took the photograph. Some times, just pointing and shooting is not sufficient. Sometimes, we have to go deeper.

This is a tutorial on how I managed to create (what I hope is) a relatively decent photograph showcasing urban decay. As with all my posts, here are the finished goods...

The photograph was taken of a backlight building, which had decayed sufficiently. However, cameras don't choose the most appropriate effect for a photograph. This is the picture SOOC.

It really does not bother me that the SOOC image does not match my vision. As long as a picture is correctly exposed and composed, it can be processed sufficiently to match my vision.

I processed this image entirely in Raw Therapee, and refrained from using the GIMP. My second take, using Raw Therapee in conjunction with the GIMP, can be found here.

Fire up Raw Therapee, and load the photograph. The first thing I do is to set the profile to "Insert Copyright". This is a custom profile I created, based on neutral. It adds EXIF copyright information, and erases any previous edits I made.

The next step is to correct exposure and increase contrast.

Followed by a high contrast tone curve

increasing the colour vibrance.

And the reverse step of decreasing the saturation. This is because I wanted to equalize saturation across the image.

The next step involves correcting the RGB curves

I make the image warmer to get rid of some of the excess blue that creeps into the image, and also to give it a more desolate feel

Finally, I add a tight crop to get rid of that distracting wire in the foreground.

And there you are! An image showcasing urban decay.

A GIMP Script for the "Retro" Look

2012-02-21T01:16:00.001-05:00

Sometimes, we tend to apply the same effect to a number of our photographs. In this case, we would like to simply have a one-click solution, instead of manually going through the same 20 step process. This is where scripting comes into the picture (pun intended).

The "Retro" look is rather popular with most people who call themselves "digital photographers", which I think is strange. You have the best digital camera, which can reproduce such perfect colours, and you still want that 70s film effect?

But it is not my place to pass moral judgement. My job is to just hand over a script that will allow you to make images look like this

using a one click filter. from an image like this

The methods I use to achieve a retro effect are similar to the methods I outlined in the tutorial on achieving a film-effect using the GIMP. However, since then, I've taken some time to refine that effect, and I present this improved version.

One point that you must remember when you script in the GIMP is that the script is essentially a programme in Scheme, a dialect of the Lisp language (no, please don't start lisping now). Essentially, you are programming in Scheme, using a GIMP API; at least, that's how I like to view it. (Remember, real FOSS users prefer to use the console for everything. So, if you are using the GIMP, I would recommend that you stop using the GUI immediately, and instead, use the script-fu console for each task. (Just Kidding. Don't worry, you can use the GUI. (If you are not yet used to the nested parenthesis, get used to them, you'll encounter lots of them while programming. (Serious!))))

My difficulty primarily is that I cannot run a tutorial on programming in Scheme, which is what you'll need if you were to write your own script. Instead, I'll invite you to find tutorials on the internet, and also direct you to the GIMP tutorials on scripting. I'll simply present the script I created and used to achieve the "Retro Effect" in the sample image.

The code is listed here:
UPDATE: The code is now available on my GitHub repository.


; Apply a retro effect to an image
; This is an example script only. Values must be changed to suit your tastes

(define (script-fu-retro-effect img 
    drawable )
 ;Create a new layer from the visible image
 (define curves-layer (car (gimp-layer-new-from-visible img img "Curves")))

 ;Insert it in the image
 (gimp-image-insert-layer img curves-layer 0 0)

 ;Modify the curves
 (gimp-curves-spline curves-layer HISTOGRAM-RED 8 #(0 0 61 41 187 207 255 255))
 (gimp-curves-spline curves-layer HISTOGRAM-GREEN 8 #(0 0 50 57 195 217 255 255))
 (gimp-curves-spline curves-layer HISTOGRAM-BLUE 4 #(0 40 255 215))

 ;Soft light layer for added contrast
 (define soft-light-layer (car (gimp-layer-new-from-drawable curves-layer img)))
 (gimp-item-set-name soft-light-layer "Soft Light")
 (gimp-layer-set-mode soft-light-layer SOFTLIGHT-MODE)
 (gimp-layer-set-opacity soft-light-layer 40)
 (gimp-image-insert-layer img soft-light-layer 0 0)
 (gimp-desaturate-full soft-light-layer DESATURATE-LUMINOSITY)

 ;Now, add vignette layer group
 (define vignette-group (car (gimp-layer-group-new img)))
 (gimp-item-set-name vignette-group "Vignette")
 (gimp-layer-set-mode vignette-group OVERLAY-MODE)
 (gimp-layer-set-opacity vignette-group 40)
 (gimp-image-insert-layer img vignette-group 0 0)

 (define img-height (car (gimp-image-height img)))
 (define img-width (car (gimp-image-width img)))
 
 ;Now, for the vignette layers
 (define vignette-white (car (gimp-layer-new img img-width img-height RGBA-IMAGE "White" 100 NORMAL-MODE)))
 (gimp-image-insert-layer img vignette-white vignette-group 0)
 (gimp-context-set-foreground '(255 255 255))
 (gimp-edit-blend vignette-white FG-TRANSPARENT-MODE NORMAL-MODE GRADIENT-RADIAL 100 0 REPEAT-NONE FALSE FALSE 1 0 TRUE (/ img-width 2) (/ img-height 2) (* img-width 1.5) (* img-height 1.5))
 (define vignette-black (car (gimp-layer-new img img-width img-height RGBA-IMAGE "Black" 100 NORMAL-MODE)))
 (gimp-image-insert-layer img vignette-black vignette-group 0)
 (gimp-context-set-foreground '(0 0 0))
 (gimp-edit-blend vignette-black FG-TRANSPARENT-MODE NORMAL-MODE GRADIENT-RADIAL 100 0 REPEAT-NONE TRUE FALSE 1 0 TRUE (/ img-width 2) (/ img-height 2) img-width img-height)
 
 ;Finally, set the tint layer
 (define tint-layer (car (gimp-layer-new img img-width img-height RGBA-IMAGE "Tint" 20 DIFFERENCE-MODE)))
 (gimp-image-insert-layer img tint-layer 0 0)
 (gimp-context-set-foreground '(5 10 150))
 (gimp-edit-fill tint-layer FOREGROUND-FILL)

 ;And flush the display so as to reflect the changes
 (gimp-displays-flush)
)

; Register the script so that GIMP may use it.
(script-fu-register "script-fu-retro-effect"
   "Retro Effect"
   "Apply a retro effect to visible image"
   "Skand Hurkat"
   "Skand Hurkat"
   "2012"
   "RGB*"
   SF-IMAGE "Image" 0
   SF-DRAWABLE "Image to apply retro effect" 0)

; This script will appear as a filter named "Retro Effect" in the Filters > Artistic block
(script-fu-menu-register "script-fu-retro-effect"
   "<Image>/Filters/Artistic")

Simply copy the script in a file named retro-effect.scm in the GIMP scripts directory. You'll find the script in Filters->Artistic when you restart the GIMP. Alternately, you can just refresh the Script-Fu scripts.

Keeping in sync with the "non-destructive editing" philosophy that I follow, this script copies the visible image onto a new layer, modifies curves, adds a soft light overlay, adds a vignette, and finally corrects colours further with a tint. Each of these layers is preserved, so if you want to change any of the layer parameters like opacity, blending mode, etc.; you can easily do so.

That's it, that's your first GIMP script. Do leave a comment if you like it, leave a comment if you modify the script when you use it, so that others can benefit from the modification. If you like this post, hit the "+1" button and make this page easier for you friends to find.

BTW, the roadside restaurants at the Old Port, Montreal offer awesome food and a great view of the streets. If only they catered more to vegetarians...

Panorama using Hugin

2012-02-07T01:59:00.000-05:00

Would you like to make a really wide angle photograph, aka a panorama? Here is an example:

Marina Bay at Night, Singapore

I made this panorama using a software called Hugin, which is a free and open source panorama stitcher based on Panorama Tools, a toolkit for making panoramas. This image was made from 5 individual images, which are below.

The individual images used for the panorama

So, download and fire up Hugin. Load the source images.

Then go to the images tab and click on create control points. This will create control points in the set of images.

Mostly, the control points are made correctly. If not, you can manually adjust them in the Control Points tab. Then, back in the assistant tab, click on align.

The images will be aligned, and a preview panorama will be shown. This may not be in the perfect orientation, so go to the move/drag tab, and adjust the panorama (Note: The move/drag tool uses arcball rotation). Then adjust the crop in the panorama. Autocrop works most of the times, but in this case, I had to make some manual adjustments.

Once you are satisfied with the preview, optimize the exposures in the optimizer tab.

Finally, stitch the panorama in the stitcher tab. That will create the panorama.

This is the final panorama.

You may observe that this panorama, while accurate, may not be as catchy. For correcting this, I simply fire up Raw Therapee (a tutorial on Raw Therapee is located here), and use it to improve the image. Mainly, I correct white balance, exposure, and increase colour vibrance.

HDR using qtpfsgui

2012-01-23T09:02:00.001-05:00

This post is a tutorial on how to create HDR images using a software called qtpfsgui aka Luminance HDR. In case you don't know what HDR is, or are doubtful about whether HDR is for you, then refer to my previous post.

The first step of course is to ensure that you have the right software. Download and install the software from the website. (Note that since the new version (2.2) released just yesterday, this tutorial is not updated to reflect that change, and uses 2.1. However, the same techniques apply; version 2.2 is much better, and I would strongly recommend that you install that version.)

HDR is useful when the dynamic range of the scene exceeds the dynamic range of your camera. For example, consider this scene

This scene exceeds the dynamic range of my camera. The dynamic range is too vast, because there is an outer façade, an inner chamber, and a chamber within that, and the only source of illumination is the sunlight outside the cave. (For those who want to know, this image was taken at Kanheri Caves at Sanjay Gandhi National Park in Borivali.) The images as taken by my camera (at 1 stop apart) are shown below.

As I own a point and shoot Panasonic DMC-FS3, I really cannot go beyond +/- 2 EV. If I was allowed to go beyond these limits, the resulting image would have shown much greater detail.

Well, so fire up Luminance HDR, click on new HDR image,

click next

and then load the bracketed images (you can drag and drop)

Select autoalign images, and select the method as Hugin align_image_stack (this uses the same align_image_stack mentioned in my tutorial on exposure fusion). Click next.

At this time, you better verify that the images have been aligned correctly.

If you are pleased with the alignment, click next. This brings up the HDR creation options.

Profile 1 works well enough for most cases, so leave it on profile 1. Now, click on finish, and the HDR image will be created.

Now, we arrive at the tone mapping operators. Refer to my previous post to understand tone mapping. Just to recall, tone mapping is a method of rendering the HDR image on an LDR output device. A good tone mapping operator should preserve local contrast while compressing the dynamic range.

What sets Luminance apart from proprietary software like Photomatix is that Luminance uses open algorithms and offers the user a choice. On the other hand, Photomatix probably uses a combination of these algorithms or its own proprietary algorithm for tone mapping. This means that Luminance ought to give you much more power upon how your image appears.

Typically, the outputs of Luminance HDR ought to be treated as layers, and each layer can be merged selectively to give best results using software like the GIMP. My preference for the tone mapping operators is Fattal and Drago. So, I select Fattal tone mapping operator, tweak the parameters as shown in this image

Make sure that you select the output image size as the maximum possible image size (equal to the size of the original images) before you click on tonemap. Operators like Fattal are based on image pyramids and give different outputs for different sizes of the resulting image.

At this point, the image looks over the top, though it retains local contrast rather well. We need to tone the image down, however. So, select the Drago tone mapping operator, and tone map the image once more.

You may feel that the Drago operator does not yield much detail.

Now, fire up the GIMP. Note that all exif data would probably have been wiped clean from the images by Luminance. So, open up a base image in GIMP (I generally prefer a middle exposure) and then open the resulting two images as layers (You saved them, I hope).

(Luminance 2.1 ignored the rotation information in the image, hence I had to rotate the two layers (Layer -> Transform -> Rotate 90 degree <direction>) before I could proceed, however, this is not necessary in Luminance 2.2.) With the Drago image above the Fattal image, add a layer mask to the Drago image, then use the bucket fill tool to fill it with grey colour (I used V=50)

Now, all that is required is to crop the resulting image, and we're done!

That's it. That's how you create your first HDR image. Now, for some caveats.

Remember that HDR is just a technique to expand the dynamic range of your camera. Do not use it for saturating every image. Use vibrance and tone-mapping operators in Raw Therapee for that purpose.

You are free to choose any tone-mapping operators in Luminance. Try the other 7 operators as well.

The reason I used a layer mask instead of merely changing the transparency of the Drago layer is that in case of pictures of the sky, the sky will turn an ugly grey with strong operators like Fattal. Hence, in such cases, you need to make the layer mask of the Drago layer completely white (opaque) over the sky. This is not possible just using transparency settings.

Use a tripod while shooting HDR. It will improve the results dramatically. I don't have one, but I wish I had a tripod.

In case you use a tripod, you can actually use one of the bracketed images to completely replace the sky in the tone-mapped images. The result will be much cleaner. In fact, try for a blend of exposure fusion and HDR-tonemapping techniques for best results