Indigo Renderer Forum

right, its just confusing, if i load the scene i can't render unless in switch image tab turn off "optimize" flag, switch back to render tab and select the render-integrators.

once bi -driectional mtl is rendering, i can go to image tab and turn on optmitiz flag (which turn on bucket rendering) without problem and render with settings with MTL with no issues. only after load the scene again i get warning and locks.

schwungsau wrote: ↑

Fri Sep 25, 2020 4:38 am

right, its just confusing, if i load the scene i can't render unless in switch image tab turn off "optimize" flag, switch back to render tab and select the render-integrators.

once bi -driectional mtl is rendering, i can go to image tab and turn on optmitiz flag (which turn on bucket rendering) without problem and render with settings with MTL with no issues. only after load the scene again i get warning and locks.

Forbidding MLT + bucket is more of a mitigation for a bug, eventually it should be possible to combine them.

You can render that way if you really want to, but the results will be wrong, and I'd recomment not doing it.

right, but rendering without mtl and denoising, then check if it faster with MTL and without denoising, and then switch back --> locked --> then you have to through jump serval loops / clicking to be able to render any pixel again....

if would more clear to have denoise option where the render settings are.... switching denoise and render integrator is a click-festival

schwungsau wrote: ↑

Fri Sep 25, 2020 6:55 am

right, but rendering without mtl and denoising, then check if it faster with MTL and without denoising, and then switch back --> locked --> then you have to through jump serval loops / clicking to be able to render any pixel again....

if would more clear to have denoise option where the render settings are.... switching denoise and render integrator is a click-festival

topaz gigapixel ai is better and fastest denoise maschine. just use it, and you'll be happy. but it's better to use topaz denoise ai for scenes with fog. indigo's denoiser has a very beta state now. i can't use it with my cheap celerons (too slow), but gigapixel works great (because of it's gpu denoiser).

wob wrote: ↑

Fri Sep 25, 2020 8:27 am

topaz gigapixel ai is better and fastest denoise maschine. just use it, and you'll be happy. but it's better to use topaz denoise ai for scenes with fog. indigo's denoiser has a very beta state now. i can't use it with my cheap celerons (too slow), but gigapixel works great (because of it's gpu denoiser).

Indigo uses OIDN, a state of the art open source denoiser developed by Intel in the last released version:
https://blenderartists.org/t/big-denois ... rs/1175114

Close to every render engine uses it now, and by using the normal and albedo pass directly from renderer its results are superior to the competitors. The AI database in OIDN is trained only on rendered content and not on all other sources like compression artifacts and camera hardware like others...
I'm interested where you have your "very beta stage" idea is rooting on.

I'm happy to offer you test image that you can run through Topaz and we compare it to Indigos denoise, are you in?!

Here is the EXR for you, give me your best shot.
I used SuperSampling of 1 to stay fair, but having higher SS raises the output quality of the denoiser to another level while keeping render time the same.
Having Higher SS values let the render denoise in the internal resolution, usually the albedo and normal pass are very clean after a short time and the extra detail they have due to the extra resolution is the key here.

Zom-B wrote: ↑

Fri Sep 25, 2020 7:13 pm

wob wrote: ↑

Fri Sep 25, 2020 8:27 am

topaz gigapixel ai is better and fastest denoise maschine. just use it, and you'll be happy. but it's better to use topaz denoise ai for scenes with fog. indigo's denoiser has a very beta state now. i can't use it with my cheap celerons (too slow), but gigapixel works great (because of it's gpu denoiser).

Indigo uses OIDN, a state of the art open source denoiser developed by Intel in the last released version:
https://blenderartists.org/t/big-denois ... rs/1175114

Close to every render engine uses it now, and by using the normal and albedo pass directly from renderer its results are superior to the competitors. The AI database in OIDN is trained only on rendered content and not on all other sources like compression artifacts and camera hardware like others...
I'm interested where you have your "very beta stage" idea is rooting on.

I'm happy to offer you test image that you can run through Topaz and we compare it to Indigos denoise, are you in?!

no, gigapixel 5 has jpg artifacts detection and blur motion detection (which will not have place in 3d renderings). and i can't see that intel denoiser is state of art denoiser. it's killing / blurring textures very often.

about beta state. indigo's denoiser is not gpu denoiser. it's cpu, and a bit little gpu denoiser. i use low class processors (pentiums and celerons) and can't use gpu render with cpu denoiser. it's nonsens to use cpu denoiser in gpu rendering modes. gigapixel ai is full gpu denoiser, based on gpu only, working fine with celerons. so i dont need to use high priced processors (why should I buy i7 high priced cpus if i will not use it in another gpu projects?? only for intel non gpu denoiser???) indigo's denoiser increases rendering time in x10-x20 on slow cpus. absolutely unusable. world #1 fastest gpu rendering engine with worst in speed denoiser...

wob wrote: ↑

Fri Sep 25, 2020 9:01 pm

no, gigapixel 5 has jpg artifacts detection and blur motion detection (which will not have place in 3d renderings).

So you don't want to show the power of your super duper denosier on this example and proof that OIDN is bad??!

wob wrote: ↑

Fri Sep 25, 2020 9:01 pm

and i can't see that intel denoiser is state of art denoiser. it's killing / blurring textures very often.

Then you are not using normal + albedo passes (optimize for denoise checkbox).
The albedo pass has all the texture information without shading, so OIDN is aware of these details and keeps them, as I explained and proofed you above, using higer SuperSampling helps here even more (I attached such a albedo pass). please please with sugar on top: show me the denoise result of your denoiser instead of this bla bla!

wob wrote: ↑

Fri Sep 25, 2020 9:01 pm

about beta state. indigo's denoiser is not gpu denoiser. it's cpu, and a bit little gpu denoiser. i use low class processors (pentiums and celerons) and can't use gpu render with cpu denoiser. it's nonsens to use cpu denoiser in gpu rendering modes.

OIDN is therefore state of the art, because it uses state of the art AI learning methodes trained only on rendering content, therefor the RESULT is the best you can get (for free).
Your crappy CPU is not a argument to make a point here!!1! CGI professionals care for quality most, OIDN is NOT a preview denoiser like NVIDIAs OptiX denoiser or others.

wob wrote: ↑

Fri Sep 25, 2020 9:01 pm

gigapixel ai is full gpu denoiser, based on gpu only, working fine with celerons. so i dont need to use high priced processors (why should I buy i7 high priced cpus if i will not use it in another gpu projects?? only for intel non gpu denoiser???) indigo's denoiser increases rendering time in x10-x20 on slow cpus. absolutely unusable. world #1 fastest gpu rendering engine with worst in speed denoiser...

Again: show me how good your denoiser is first! CGI Artists dont need a fast but low quality PostPro denoiser that aims for photo and Video content.

Are you on a retired mining rig with bunch of GPU power but a low power CPU?
Simply denoise only on the need of rendering, have the denoise checkbox off until you want to denoise your finished image.
If you use Indigo_console.exe denosing is only applied after halt parameter is triggered and before saving the output.

Are you on a retired mining rig with bunch of GPU power but a low power CPU?
Simply denoise only on the need of rendering, have the denoise checkbox off until you want to denoise your finished image.
If you use Indigo_console.exe denosing is only applied after halt parameter is triggered and before saving the output.
[/quote]

i have two PCs with intel core i7-9600/9700 + RTX 20.. to work in cinema4d / blender and cloud organized gpu rig 1070/ti with ~5200 OctaneBench 2020.1 power based on slow cpus. and it's possible to use all fuctions of rendering / denoising only with gpu power. i'm not pro-3d artist and not a rendering pro and not a rendering-power seller so I won't invest extra money in something I won't use (high-priced cpus).

but i know rendering / denoising engines very deeply. now i use only gigapixel ai and davinci resolve temporal denoisers. indigo's intel is not suitable for me because of it's not a gpu denoiser. and i think, mining is not shameful ;)

I think the denier works pretty good and superfast. topaz has problem it does not work on animation. the topaz tools are pretty awesome for upres'ing.

@wob
But why can't you keep using those gpu denoisers, do they cease to work when indigo's working?

schwungsau wrote: ↑

Sun Sep 27, 2020 6:42 am

I think the denier works pretty good and superfast. topaz has problem it does not work on animation. the topaz tools are pretty awesome for upres'ing.

topaz has fixed their famous batch processing bug in 5.0 version of gigapixel.

Still waiting on that comparison :D

PS. It's "MLT" (Metropolis Light Transport), not "MTL".

mpro31 wrote: ↑

Thu Sep 24, 2020 1:29 am

Hi OnoSendai,
floating licence does not work in new version
any suggestions?

Hi,
It works fine in my testing.
You will have to supply further details, logs etc..
You can post here or email us at support@indigorenderer.com.

Can you send "C:\Users\xx\AppData\Roaming\Indigo Renderer\log.txt" and
"C:\Users\xx\AppData\Roaming\Indigo Renderer\network_manager_log.txt"

and a description of the steps you took to try and use floating licences?

i just tested Indigo 4.4.14 indigo in network mode. indigo 4.4.14 crashes in network render mode, when i start in with 2 nvidia cards. this problem appeared in 4.4.10 and doesnt fixed up to this day... works only with single nvidia card. standalone master works wine with 2 cards.