OTOY Forums

Refracty wrote:... or you rotate the source picture in Photoshop

I did that too

But such hacks don't work if you need to rotate the image by 15 degress, or translate it by 1 cm, or scale it only by X... We need proper tools in Octane.

1.) We need to be able to pick a coordinate source for some texture (UV, world global, object origin...)

2.) We need to be able to pick a projection for some texture (cube, cylinder, sphere, flat (when source is UV),...)

3.) We need to be able to fully transform these coordinates, with a transform node (scale XYZ, translate XYZ, rotate XYZ)

4.) In case of instances we need to be further able to diversify texture mapping by some object (instance) specific data, like: randomly generated for each instance, by it's position or by some other arbitrary data read from the .cvs (supplied by the modeling app)

Matej thats the Point. This is EXACTLY what i mean. You nailed it. My english is not so good, but you really got the point.

Who needs lens flares??? This is so much more important. I can do Lensflares or a Glowing effect within seconds in Photoshop. Mapping textures like this takes HOURS if you have a huge scene.

I wish it would be like in the keyshot picture i post here now - soory it is german, but i am sure you know what it says. if you dont understand the dialog, i try to download the english keyshot version and post it again, but it really should be clear, cause it has ALL in it matej, what you wrote above and it is SO simple to use!
Kind regards Chris!

well, in defense of the lens flares, all the post effects are being developed by someone who specializes in that stuff and they got him on the team to do post effects. So we are not missing out on other things by having these post effects.

I've pretty much assumed that these tools were on the way. I hope very soon. As well as displacement. At any rate, GPU rendering is a whole new ball game, and there is much development that needs to be done. Don't know if I am being too kind, but programming a GPU is a different animal than programming a CPU to do stuff. So I would expect that many tools will have to wait as the software technologies get built from the ground up. As a programmer myself (but not a GPU programmer) I assume that the core issues that needed to be solved were basic GPU rendering algorithms and specifying how the GUI would behave, that once built would be the foundation for tools. Give the tools a chance to evolve out of that foundation.

treddie wrote:At any rate, GPU rendering is a whole new ball game, and there is much development that needs to be done.

The tools discussed here don't fall into this category (ie. needs time to do R&D for GPU implementation). Cycles runs on the GPU and has all sorts of texture manipulating / mapping options. AFAIK Keyshot also runs on the GPU. Direct Octane competitors: Arion, Vray-RT - I don't believe for a second that they lack such fundamental functionality as coordinate transforms or auto projections (but someone who is familiar with these engines, can prove me wrong)

Could be right. I have to admit my programming experience is entirely CPU, although I have talked to a number of GPU programmers and they mentioned that it's harder to do than on a CPU. Also, I don't know whether or not they have to build all of their libraries from scratch, or whether they are using entirely off-the-shelf libraries for CUDA that include standard tools.

Another possibility is what someone on the Octane Team had said a long time ago (or that I INTERPRETED them to say), that the standalone renderer was not meant to be a "studio" type of environment...just a renderer. Whether or not my interpretation of his comments is correct, I feel that it is a moot point...The market will force Otoy to move to a studio style with all of the requisite tools.

AFAIK Keyshot also runs on the GPU

Matej keyshot is not using GPU for rendering. The gpu just helps when setting up the scene, but does not calculate shadows, caustics and so on. Arion uses CPU and GPU and yes it is having all the possibilities like Keyshot has. So - it is possible and implemented in them, but i dont know how complicated it is to implement those features. I normaly do not complain - but beta is over - and it is not ANY feature - this is the basic of the basic...............of the basic. This is for me the same as i would not be able to put some emitters in a scene. could you imagine to work without emitters? no....and this is (for me) the same.

Imagine making a Tire and add "Goodyear" on it. Done in keyshot in under 30 seconds. Try it in Octane..... No - this has to be in Octane in the next release or i am gone until it is implemented. I need it - its a must have for my work and time is money here.

Kind regards Chris!

@treddie, apparently some algorithms are "impossible" to implement on the GPU (like bidirectional pathtracing), or at-least their development is dependent on how CUDA development progresses.

But I believe this is not the case of the features mentioned here. We already have a transform node that works on procedural noises mapping space, we just need it to work also on image texture UV space.

Having powerful tools for mapping and re-using your textures is crucial, because we are working in a very RAM-limited environment, so this should be top priority on the TODO list. Having different coordinate sources (like global or object local, not only UV) could already help to make our scenes more diverse, without increasing RAM usage too much. For example; placing decals on walls - right now you can't do it satisfactory (= without using huge amounts of RAM) in Octane because all you have to work with is one single UV map for all the textures that belong to some material / object. In Cycles: you map your base wall texture in global space or object space and your decal textures in UV space. You can fully transform each space (per texture!) with a mapping node. All this gives you flexibility to build & diversify your materials without huge costs in RAM.

These Octane limitations were understandable in early beta stage, but we are now in "final" 1.0. We are not asking to invent new never-seen-before algorithms here, but basic functionality that every renderer out there has.

also, shading and texturing improvements were already discussed with RS a long time ago: see here (admittedly some of the stuff mentioned there was already delivered, but "texturing workflow" still badly lacks functionality)

Matej - again you nailed the point haha! Funny to read the thread with final release. So many said texture, UV.......and now - in the final it is not implemented. Wow.

So....if we talk about missing features......i habe another Question....: Maybe this will work in Octane......but i dont think so - cause it would be usefull.........

My Question is: in Arion for example you can pick a material and simply apply the same material to another objekt with ALT+left Mousclick. So its very easy and fast to apply Materials. Is there ANY way to work like that in octane? Would be so great! Please tell me .......YES

Kind regards Chris!

Well, one thing that would help, would be if Otoy would share with us their blueprint for the immediate future...Like their roadmap up to v3. They don't have to reveal their secret stuff, just the common stuff we all need and when we can expect it.