topic Re: Texmap comparing in 3ds Max Programming Forum

Texmap comparing

Anonymous — Tue, 04 Jul 2017 10:48:08 GMT

Some time ago I've posted question (and received answer) about rendering bitmaps at C++ SDK:

https://forums.autodesk.com/t5/3ds-max-programming/exporting-resulting-bitmaptex/m-p/6987392

It was very helpful, but since then I've met new problem: how to avoid duplication of identical (or, better, very similar) textures? Thing is, rendering bitmaps is not very fast (though, not very slow) operation.

Consider we have big scene where user used few textures at many maps and modifiers, such that he has, actually, "new" textures.And consider that we want to export it. There would be

1) full duplicates

2) situations where we can just receive image from drive despite of rendering.

Does anyone know, if there is way to compare texmaps?

Re: Texmap comparing

denisT.MaxDoctor — Tue, 04 Jul 2017 20:35:34 GMT

By what criteria do you want to compare? Or is this the crux of the matter?

Re: Texmap comparing

Anonymous — Wed, 05 Jul 2017 07:34:34 GMT

Well, I think there would be not that much of problem to compare if I would have rendered images. It could be even done by external tools like ImageMagick's Compare with criteria on some metric (absolute error, mean error, direct comparing pixel by pixel, hashing with comparison on collisions etc.).

Thing is, I would be interested how could I find out that not rendered (yet) by Max's C++ SDK textures are equal (if it's even possible) or close to each other?

I quite understand that this question is out of plane "How Max can do it" but rather in plane "How Max's programmer can resolve this problem on his own".

Though, still. Right now I'm looking if in material in needed map (diffuse, for example) some bitmap with external image is linked directly and just using this image and bitmap's tiles|offsets, or there is a chain (that has external image in it) and I should render it.

Nevertheless, I would like to look if some map was rendered before as I've often met situations when in different materials there are same chain's of modifiers for bitmap (like 'Color correction', for instance) with same parameters and, actually, we are rendering this image twice (and storing it twice). That already made problem as once I received 2Gb exported archive instead of 600Mb from scene (not to mention that rendering bitmaps on client's machine increased time of exporting drastically and that it needs much of space on client's machine in process).

Thus, maybe, my question should be "Is there any way that modifiers on some image(-s) are identical?"

Well, probably my problem is quite outstretched and can't be resolved easily. Thereby, I'm just looking for advice what could I do. And if some possible solution may come to you.

Offtop: Are you almost the only one user who assists others here? As I've met little of others in my posts aside of you.

Re: Texmap comparing

Anonymous — Wed, 05 Jul 2017 08:15:23 GMT

Also, thought just right now. When I debugged yesterday my code, I found that rendering was faster than writing on drive. I think, it could be reason too.

How do you think, in this case should I make lesser variants of images and hash them for fast compare? And after all export render big variant of image.

P.S. Even if I think it's dumb way to do things. But right now I'm going to check time amount.

Re: Texmap comparing

Anonymous — Wed, 05 Jul 2017 09:21:59 GMT

95.6 seconds of render time against 454 seconds of writing bitmaps in output for 134 images with width/height 1k-2k mostly.

CPU Intel Core i7-7770 3.60GHz, 64 Gb RAM, NVIDIA GeForce GTX 1080.

Hashing smaller images question is still actual.