Some 3d model marketplaces start to allow (or push) sellers to give their 3D models to AI training. I have no more detailed info on the matter so my assumptions are –
they hope to get a huge chunk of data for AI people to be able to generate 3D models. If it is a success it’s a win for AI tool providers, A short term gain to 3D marketplace that gives models
A win for a customer i suppose, and a loss for 3D model creator / artist. (as well as marketplace in long term)

So looking from an individual 3D model seller perspective – we are doomed. And even one time 3d model sold for AI people is just the last sale till there is no market anymore.

So what could be a way forward? Change profession. 

But other than that… imagine a 3D AI model generator that is not universal (trained on millions of models owned by big data corporation) But it’s a tool that only spits out different versions of specific models. So a 3d model works like a style “preset” – if you like a model you get something very similar but unique as a buyer.

Now can an individual do this realistically? probably not. But imagine a 3D marketplace that feels a threat of AI generators and that for some reason doesn’t steal all models and become an AI tool manufacturer…. imagine it would develop such a tool. That would be not as universal for a buyer, but it would guarantee consistency, quality and still retain uniqueness – in a sense that as today, users find a model (or models) from a seller they like and the site generates a version specifically for them. A unique version.Might be this tool would be even more useful if a buyer could add to cart models from different sellers, and the site would generate variations based on the set of models.

just a thought.

How is it better (to a buyer) compared to a big universal generator? Well… hard to tell. one assumption – more ethical. also. it may or may not be better in a sense that

it potentially would work much better in a sense that you (as a buyer) would get a much more precise and consistent style based on the model you selected. Now users need to do a lot with prompt engineering.

just a thought.

Planing part II using mobile so only vertical view for now. Restrictions I have to work with with limited set of models. Flow chat drawing on mobile for puzzles in games. All in all this is a dev log that might be interesting for people working with 3d graphics and planing to step in to real time engines as well as doing some photogrammetry and all of that in Virtual Reality. So its still planing part, and here also show some snippets from my previous VR environment and some workflows very briefly, just to give you a taste of whats coming.
here is a video from my first environment – https://www.youtube.com/watch?v=t7SRu…
and the environment itself – https://steamcommunity.com/sharedfile…
and joggling in vr with carrots and skulls just like in hl alyx!

So in the third part I am doing retopo of a scanned model in 3D coat. then splitting the result for better uv shall / island distribution to maximize texture space. And last thing – sins steamVR / hammer / source 2 (?) doesn’t support tessellation / subdivision, I am “baking tessellation” to the mesh using height map that i have from baking my original model.

In the forth part I am testing if it is worth to split lower part of tower to 5 separate models each with 4k texture and sort of a displacement map baked directly in to the mesh sins there is no tessellation in hammer editor. (At least steam VR version of it). Also accidentally recorded a hammer crash, that just happens each ten mouse clicks. Well to be honest might be each 20 mouse clicks :) so save regularly and pray..

Strange problems with hammer editor
normal maps or smoothing groups (max terminology) soft edges (maya terminology) phong breaks ? (in cinema 4d) shade smooth ? (blender)

so for some reason importing obj file to hammer brakes smoothing groups and only way to fix it i found is selecting all edges on object in hammer and setting normals to smooth. in general more detailed walls do look better then just flat ones, so i will use quite a dense mesh for my tower. its around 15 k – 20 k for ground, and all 4 walls and a dome as one model.

One small update on normal maps. Sometimes you might end up in a situation where normal maps are not as punchy not as strong as you might want them to be. But you do not want to spend time doing things again. so i found this sort of cheating method. it kind of works. Basically it makes your existing normal map to have more effect. sort of. all you need to do is duplicate the image in 2d image editing software and modify its “levels” to be specific only blue channel of an image. And you use this as overlay blending mode on top of original image, save it and here you go. I should do more A B tests to see is it worth it, but at least it doesn’t make it worst or it seems so :) go ahead try it out and report me back if it works for you as well.

Continuing my development of photo realistic 3d scanned VR environment for steam vr in 3d studio max. Sins I have spitted upper and lower part of my tower walls to have better texture density i end up having unwanted seams that i plan to hide by making beams (big wooden things that you find in ld architecture). Here I only make a box, add some edges, some noise some chamfer and make uv’s.
Main take away – dont model too much detail, and see at what point its best to do uv unfolding. You make it later in process its harder to unfold. You make it too early and you might destroy something. Adding some chamfers, noise’s and stuff could be done after unfold is done and uv’s are done.

In this episode we will continue with VR environment (with photogrammetry assets) devlog. In last part we created beams (jolts?) and made their uv’s. Now lets draw textures using texture projection. I found that 3D coat lets us use one texture to be simultaneously painted in normal map and color textures – that is super neat! So i decided to draw some burned wood textures that i have prepared in photoshop by mixing up some reference photos. Drawn two versions of wood texture – regular and burned. Made a mask in photoshop to mix up these textures. And load it up to engine.

photorealistic 3D scanned models

why arguably one of the biggest 3D scanned model providers (quixsel) slashing its prices or making them free in UE asset store is a good thing for smaller photogrammetry asset creators like me.

I would imagine only big AAA game dev. studios can actually produce artwork that can look good together with photorealistic scanned models. And such studios, i imagine rarely buys assets, they make them in house. So i imagine indie devs, who actually sometimes buy models in asset store, can not make such quality models by themselves with their indie budgets. Having a big scanned object library go for free, entices smaller and indie devs to try AAA graphics and they become potential buyers of other scanned models. Basically, more people could now be my customers.

..or am I just looking at the world through “rose-tinted” glasses, doping prices will lead others to drop prices and it all goes to… you  know where…

“hero” models or environment props?

As a 3D model sellers in marketplaces, are we complicit in asset flipping? Asset flipping is something we despise in game industry. But same approach in arch viz and product design does not make sense. If well made, a car of a same model will look the same no mater the artist. In games though we are creating unique worlds and asset flipping becomes much more of a sour spot if it stands out and gets noticed. And there are some community members who are not only allergic to asset flipping but vocal as well.  And here, i argue, photogrammetry (or any other 3D scanning method) comes in. So if we take a tree, a stone, a brick wall…

unique or generic?

As long as they are not main characters and as long as we strive for that AAA – photo realistic look, they all will just look real. So i think asset flipping in realistic graphics, for non “hero” props becomes way more forgiving. The model that is somewhat stylized is quite recognizable, it has certain style. 3D scanned models are most realistic and no one calls photorealism a style that someone used, therefore, all others using it are “copy cats”. So doing asset flipping with such models should be less noticeable and therefore again should broaden appeal of buying 3D scanned models for game dev. Who would recognize that a same tree was shot in 5 different movies? it just looks real, it doesn’t carry a “style” that would be recognized to it.

Enter a caption

So this is quite old well. I get drinking water from it when i am living in my village. Its our neighbors well. So as you can see its main construction component is these… concrete barrels that used to be commonly used in all wells i know of similar time period. Not that exiting. But nice part is all this wooden part to protect well from rain, to store a bucket and so on. Also clutter around well is exiting and interesting for me at least. All you need is a rope and a bucket and you can use the well for any game dev situation it can be both decorative and functional piece.

You can get it directly from sketchfab (3D image above) or on CG trader water well 3d model.

https://skfb.ly/6zHT6
And here is the scene where i used this model:

https://skfb.ly/6zAvR

So inspired by result i tried to make some grass scans. And here is the result – it turned out pretty good i would say. The reason i chose plantain is that it has big leafs – easier retopology and good ground coverage. so compared to simple grass you need less polygons to have convincing foliage on a ground.

https://skfb.ly/6zKLY

You can get this model in sketchfab but as well you can get it on cgtrader – here – with a 30 percent discount now!

So on Cgtrader: Door I and Door II

So that was Jesus. What about Mary? Here she is:

Ok, the bubbles and twisting is not part of the model. But now she looks cooler, doesn’t she? ah dono. So Again quite a common cast sculpture commonly found in cemeteries. This one has auto retopo but in 3 levels. the Mary herself has most polygons, the rock has less while cross has a least number of polys. Again 4k textures.

But.. can you actually see anything in this gif.. not so sure. so here is a proper picture of this 3D model as well:

and if you are game developer and you work with unity -why not to get all graveyard models i have in one pack on unity store? click on image below:

  comes with prefabs, colliders and LOD’s

These Models where made in two cemeteries in Vilnius (Wilno) in Lithuania – Bernardinų cemetery and Rasų cemetery. Most graves are from the end of 19th century and most inscriptions are in Polish, sins that was quite a Polish city at the time. All graves are broken, rusty and run down. So if you are looking for some background objects for your new game that might be in horror genre, or historical, or just takes place in Europe you might be interested in these models. Actually models are low poly but still have quite a bit of detail and they might be used not only as a background objects. Poly counts are in 2-4k range depending on model. All have diffuse (Albedo) and normal maps. Some have Ambient Occlusion, Height or Bump maps. Textures are mostly in 4k resolution. You probably can use these models not only for real time graphics but for visualizations or film as well.

Models are sold as separate gravestones/crosses, as collection of a particular graveyard or all in one.

Here are links to skechfab depicting two sets – from both grave yards.

Models from “Bernardine” Cemetery:

Models from “Rasu” cemetery:

And here is All in one collection on Cgtrader:   all gravestones, crosses and thumb stones

Rest of the models: All other models