i dont think Optix is better for animations, it might be more temporally consistant but it looks way worse overall. i used to use it thinking it better, but switching back i realized why OpenInage is default
interesting, i've had quite good results with both. I think the temporal consistency you mentioned helps out a lot for scenes with a lot of movement, but for others OpenImage could definitely compete with it 😁
If your using the Optix denoiser shouldn't you then switch your system render device to Optix? Does this thin slow down the actual render before denoising if your graphics card benefits from cuda cores? Feel like there should be a way for it to switch from CUDA for the render and then Optix for the denoiser or does this already happen? Not sure if I am competelly misunderstanding how to denoisers work.
i've heard that jpeg and png, if they are both the same bit depth, end up as the same size once loaded into your VRAM since the GPU has to use an uncompressed image. am i mistaken?
I didn't know that but I'll do a bit more research on image types with regards to rendering and RAM, using jpg definitely makes the scene "lighter" while working, but that might change when you hit the render button 🤷♂ time for me to google! 😃
I don't think this video is useful at all. Looking up "how to render faster" on youtube, everything you've said has already been said over and over, nothing new came out. And Optix is worse every time btw, so that's a mistake. Didn't even mention temporal denoising or more advanced techniques. I'm sorry but i just don't see the value of this video.
i really appreciate the honesty. I've honestly never done a strict A/B comparison of the two denoisers but I've used them both and had good results from each, I'll definitely read up a bit more on the two versions and use that to improve future videos would you be against going into a little bit more detail about the "more advanced techniques" you're referring to? I'd love to read up on them so I can make better videos in the future 😀 again I appreciate the feedback! cheers
Not useful at all is a bit harsh. It is useful to those completely new to understanding. i understand it can get repetitive. the point is to pack it in a short video so you dont have to watch a 20 minute to cover the same stuff he has. He's right, people dont still switch GPU from preferences.
@@scenefiller more advanced techniques: - Various forms of temporal denoising, especially the node based ones in the compositor using motion vectors. Also there are a couple addons that offer this feature ( i don't remember the name) - Baking lightmaps for static scenes or rough materials. For example, if you have an animation inside a house, and the lighting doesn't change, you can bake the cycles lighting to a texture and use it on an emission shader on any object using a very rough material, like dry wood or a concrete wall. This way, you don't have to calculate lighting for every frame when it's always the same. - Using the Holdout and Indirect Only modes on objects in the scene. This is fairly specific and i can't really explain it in a comment, but ypu basically exclude some objects from the scene and render them later at different settings. - baking complex procedural materials to textures - optimizing certain shaders like for glass objects, using the shadow output of the light path node and using it as a factor to mix a transparent bsdf shader with a transmission one - Using compositing tricks with the depth pass to fake fog, depth of field and parallax (ian hubert has a video on it if i recall correctly that shows how you can turn a single frame in a 3d scene with the depth pass) - Using EEVEE to render fog and volume and then compositing it onto the Cycles render Etc. UA-cam is already full of 2 minute videos covering the basics. Blender noobs are already well fed. Most advanced tutorials are overy long, tedious or just badly recorded, so it'd be great if someone collected all the more advanced things to optimize rendering times into a well recorded playlist of high quality videos
@@ShoryYTP I really appreciate the followup there, I'll look into each point and learn a bit more about them and hopefully future videos can be a bit more in-depth 😄 I definitely agree with you that there's a content gap for some of the more advanced topics (probably because they're advanced), so it's not a bad idea for me to try and make some of those concepts a bit more reachable for people (of course I'll still be doing videos on the basics as well) thanks again!
you earned a sub. fast, concise, accurate information. thank u so much 🙏
Bro your tips have helped me a lot
i dont think Optix is better for animations, it might be more temporally consistant but it looks way worse overall. i used to use it thinking it better, but switching back i realized why OpenInage is default
interesting, i've had quite good results with both. I think the temporal consistency you mentioned helps out a lot for scenes with a lot of movement, but for others OpenImage could definitely compete with it 😁
If your using the Optix denoiser shouldn't you then switch your system render device to Optix? Does this thin slow down the actual render before denoising if your graphics card benefits from cuda cores? Feel like there should be a way for it to switch from CUDA for the render and then Optix for the denoiser or does this already happen? Not sure if I am competelly misunderstanding how to denoisers work.
i'm not sure actually, I'll dig a bit and see if there's any benefit or reason to use optix alongside the optix denoiser 🙂
i've heard that jpeg and png, if they are both the same bit depth, end up as the same size once loaded into your VRAM since the GPU has to use an uncompressed image. am i mistaken?
I didn't know that but I'll do a bit more research on image types with regards to rendering and RAM, using jpg definitely makes the scene "lighter" while working, but that might change when you hit the render button 🤷♂
time for me to google! 😃
I don't think this video is useful at all. Looking up "how to render faster" on youtube, everything you've said has already been said over and over, nothing new came out.
And Optix is worse every time btw, so that's a mistake.
Didn't even mention temporal denoising or more advanced techniques.
I'm sorry but i just don't see the value of this video.
i really appreciate the honesty. I've honestly never done a strict A/B comparison of the two denoisers but I've used them both and had good results from each, I'll definitely read up a bit more on the two versions and use that to improve future videos
would you be against going into a little bit more detail about the "more advanced techniques" you're referring to? I'd love to read up on them so I can make better videos in the future 😀
again I appreciate the feedback! cheers
Not useful at all is a bit harsh. It is useful to those completely new to understanding. i understand it can get repetitive. the point is to pack it in a short video so you dont have to watch a 20 minute to cover the same stuff he has. He's right, people dont still switch GPU from preferences.
@@scenefiller more advanced techniques:
- Various forms of temporal denoising, especially the node based ones in the compositor using motion vectors. Also there are a couple addons that offer this feature ( i don't remember the name)
- Baking lightmaps for static scenes or rough materials. For example, if you have an animation inside a house, and the lighting doesn't change, you can bake the cycles lighting to a texture and use it on an emission shader on any object using a very rough material, like dry wood or a concrete wall. This way, you don't have to calculate lighting for every frame when it's always the same.
- Using the Holdout and Indirect Only modes on objects in the scene. This is fairly specific and i can't really explain it in a comment, but ypu basically exclude some objects from the scene and render them later at different settings.
- baking complex procedural materials to textures
- optimizing certain shaders like for glass objects, using the shadow output of the light path node and using it as a factor to mix a transparent bsdf shader with a transmission one
- Using compositing tricks with the depth pass to fake fog, depth of field and parallax (ian hubert has a video on it if i recall correctly that shows how you can turn a single frame in a 3d scene with the depth pass)
- Using EEVEE to render fog and volume and then compositing it onto the Cycles render
Etc.
UA-cam is already full of 2 minute videos covering the basics. Blender noobs are already well fed. Most advanced tutorials are overy long, tedious or just badly recorded, so it'd be great if someone collected all the more advanced things to optimize rendering times into a well recorded playlist of high quality videos
@@ShoryYTP I really appreciate the followup there, I'll look into each point and learn a bit more about them and hopefully future videos can be a bit more in-depth 😄 I definitely agree with you that there's a content gap for some of the more advanced topics (probably because they're advanced), so it's not a bad idea for me to try and make some of those concepts a bit more reachable for people (of course I'll still be doing videos on the basics as well)
thanks again!