Standalone not in great shape

[psanitra]

I`m trying to use standalone a lot past half year.The experience been quite painful.

My understanding is, that standalone should provide rendering independently from host app or OS and provide very same result each time. It should not matter if i run it on 10 local machines in the office with Win 10, or on GCP/AWS Linux machines. Seems like this is not the case right now.

- Output Render settings in Max vs what is written in vrscene file does not match.
Namely the settings to save deep exr with separate RE passes in separate exr. I need to edit vrscene file in text editor to fix it. Adding relements_separateFolders=1; Adding OutputDeepWriter vrayOutputDeepWriter plugin, bits_per_channel=32; etc

-There is no way to properly to set 16 vs 32bit exr to individual output files(RE,DEEP etc). It`s all 16 or all 32bit. This goes for all vray hostapps, standalone etc. This is specially painful when rendering on external farm, or cloud. It`s huge waste of storage and upload/download bandwidth(yes,for vray cloud too)

-For bit depths, you made a band-aid, in form of force_32_bit_output=1; that you added to VraySamplerInfo or other world related passes(zdepth). But that`s half broken. You forgot to write EnableDeepOutput=0/1; into the vrscene file. So instead having just the deep data in deep exr, all positions RE are saved too(all Samplers RE,zdepth). More headache to find out what`s wrong, no proper documentation for vrscene file on help docs how to fix it. Trial error..., 300Mb file, instead of 100MB

-As far as visual output, on latest update 2, it seems like flipping axis on vray normal maps is not being written properly to vrscene file and differently in standalone. They just don`t match, one is flipped other not.

-Noise map(noise, cellular) seems to have different seed, when i render local from max(Win10) and render from standalone on GCP Linux. Same 4.2 versions.


-I would love to see the bith depths issue being resolved for good. It`s been bothering users for years(All aps)
When render is finished all the data are present VFB, and they don`t have bit depth restriction, it`s just a matter of what to write, where and how, I don`t see how saving image file could a problem...., just let us decide what is 16 and what 32 , for example by adding something like "bit_depth_output=16/32;" and expose the controls per RE in host apps. Allow us to have and option to save Crypto to multilayer exr on its own.

- Ressurect the vrscene editor project please, i saw some remnants when googling why did you abandon it?
Why we have to edit text files in 2019? or having to write custom GUIs apps to edit scenes? https://vrscenegui.babylondreams.de/index.html#home

- Thinkbox Deadline cannot deal with standalone vrscene file properly eighter. Simply because having command like arguments to vray.exe is not enough, there is only that much one can do with it, and it will always be broken/lag behind, because new versions/development cycle.
Another options with deadline is to parse vrscene files with python before render time. That`s not very user friendly eighter. You need people to code it, machines to run it, it`s just extra time and $. It`s completely unnecessary.


We need to have vrscenes exported properly and be able to do it on our own effectively. What we need is the V-ray Scene Editor with GUI indeed.

-to access all the render parameters(resolution, sampling, engine type etc)

-paths, list them , repath the maps, set the output paths, for example output relative to vrscene file location, or to enviroment variable, for example $RENDERSHARE\JobXYZ\Shot010\

- RE depths, List all elements, adjust paths/names, bit depths.

- Ability to add vray plugins, for example Adding OutputDeepWriter

- take the attached image for example of other people that tried..




We just want to render, that`s what V-ray is for right. If we pay for each single render node licence, at least let us render. Make it easy for for us to control render settings, to adjust for each pipeline individual needs. Make Standalone really what it should be - a standalone.

thank you.

[vlado]

See some notes below. I should also point out that whenever you have issues, you can ask - most of the problems you outline below have solutions or workarounds.

- Output Render settings in Max vs what is written in vrscene file does not match.
Namely the settings to save deep exr with separate RE passes in separate exr. I need to edit vrscene file in text editor to fix it. Adding relements_separateFolders=1; Adding OutputDeepWriter vrayOutputDeepWriter plugin, bits_per_channel=32; etc

You can override any .vrscene parameter on the command line without editing the .vrscene file manually. You can use the -parameterOverride command-line flag for this purpose. It is true that you can't add a new plugin like OutputDeepWriter though. It should not be too hard to implement, if you let us know that it is an issue.

-There is no way to properly to set 16 vs 32bit exr to individual output files(RE,DEEP etc). It`s all 16 or all 32bit. This goes for all vray hostapps, standalone etc. This is specially painful when rendering on external farm, or cloud. It`s huge waste of storage and upload/download bandwidth(yes,for vray cloud too)

There should be no need to have to set the bit depth per channel; setting your main OpenEXR output to 16 bits should be enough; V-Ray knows that some render elements like Z-Depth, point position and a few others need to be written in 32-bit always, regardless of the chosen setting, but the rest will be written as 16-bit. The ExtraTex render element has manual control of this because we don't know what's in it. Other than that, if V-Ray does something unexpected here, let us know and we will look into it.

For bit depths, you made a band-aid, in form of force_32_bit_output=1; that you added to VraySamplerInfo or other world related passes(zdepth). But that`s half broken. You forgot to write EnableDeepOutput=0/1; into the vrscene file. So instead having just the deep data in deep exr, all positions RE are saved too(all Samplers RE,zdepth). More headache to find out what`s wrong, no proper documentation for vrscene file on help docs how to fix it. Trial error..., 300Mb file, instead of 100MB

This is super easy to fix if you let us know that there is an issue like that. I will try to include it in the coming hotfix for Update 2.

-As far as visual output, on latest update 2, it seems like flipping axis on vray normal maps is not being written properly to vrscene file and differently in standalone. They just don`t match, one is flipped other not.

Thanks for the pointer; will look into it and fix it.

-Noise map(noise, cellular) seems to have different seed, when i render local from max(Win10) and render from standalone on GCP Linux. Same 4.2 versions.

Hm, this should not be the case. Will try to reproduce it here but it will be helpful if you have some examples.

-I would love to see the bith depths issue being resolved for good. It`s been bothering users for years(All aps)
When render is finished all the data are present VFB, and they don`t have bit depth restriction, it`s just a matter of what to write, where and how, I don`t see how saving image file could a problem...., just let us decide what is 16 and what 32 , for example by adding something like "bit_depth_output=16/32;" and expose the controls per RE in host apps.

I would like to avoid providing manual control over this; like I said V-Ray should do the right thing by default on its own and will write the channels with the correct bit depth depending on what they represent. If it doesn't let me know and we'll see what can be done about it.

Allow us to have and option to save Crypto to multilayer exr on its own.

This is planned at some point.

- Ressurect the vrscene editor project please, i saw some remnants when googling why did you abandon it?
Why we have to edit text files in 2019? or having to write custom GUIs apps to edit scenes? https://vrscenegui.babylondreams.de/index.html#home

There should be no need for you to edit .vrscene files. Any parameter can be overwridden on the command line using the -parameterOverride command-line option. Adding content to the .vrscene (like a deep output writer) should not be too hard to add if needed.

- Thinkbox Deadline cannot deal with standalone vrscene file properly eighter. Simply because having command like arguments to vray.exe is not enough, there is only that much one can do with it, and it will always be broken/lag behind, because new versions/development cycle.
Another options with deadline is to parse vrscene files with python before render time. That`s not very user friendly eighter. You need people to code it, machines to run it, it`s just extra time and $. It`s completely unnecessary.

There should be no need to do anything with .vrscene files; our goal is to be able to do everything that you need on the command line. If there is something you can't do, let us know and we can certainly look into implementing it.

We just want to render, that`s what V-ray is for right. If we pay for each single render node licence, at least let us render. Make it easy for for us to control render settings, to adjust for each pipeline individual needs. Make Standalone really what it should be - a standalone.

One thing I would like to point out is that the 3ds Max <-> V-Ray Standalone workflow is work in progress and we would be glad to work with you on solving any issues you are having. Other integrations like V-Ray for Maya, do not have many of these issues because they were built from the ground up to work with the standalone.

Best regards,
Vlado

[psanitra]

I`ll prepare some example max scene and vrscene exported scenes, including not matching renders exr etc. I`ll let you know when the package is uploaded on your FTP.

Currently, to submit and adjust a standalone job to render is not simple and straight forward experience. It does not need to be like that. Hopefully you will find time yourself to go trough the process of rendering these files correctly, approach this from the user end, and experience what i did. Maybe then my post will make more sense.

[vlado]

It would be great if you can prepare some examples - thanks!

Best regards,
Vlado

[psanitra]

vlado , just checking if you got the files, i mailed directly and archives are on your FTP.

[vlado]

Yes, I got the files, many thanks! Will get to this in the next couple of days.

Best regards,
Vlado

[Syu_z]

psanitra A little off topic, but your sig for psanitra.com, it repeatedly refresh on my ipad due to problem in the webpage, resulting in crash of the said page.

Is there any resources to look at on how to denoise with standalone effectively? I am working in a company that still uses vray 3.6, and doesn't use inbuilt denoiser because the render elements are not denoised, so i would like to explore this avenue.

[hypotheticalinc]

I'm having a similar issue - not sure if it's already covered by what you all have been talking about. In Vray 3.x I could add deep image data just by overriding img_deepFile and img_rawFile to both be true. In Vray Next this doesn't seem to be enough as I'm not getting deep images. Do I need to add OutputDeepWriter manually / scripted for now until a new build is released to add that automatically?