-
Notifications
You must be signed in to change notification settings - Fork 49
AY-7315_Extract Review and OIIO transcode failing to transcode media blocking publishes #989
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
Linking this prototype I once did to not raise an error but instead log a warning to at least allow continuing to publish: #988 |
I kinda do get grasp of the problematics, but I don't know full purpose of The proper solution would probably be, that maya would pass the information about output renders to publish job, which would require massive refactor of deadline addon and all host integration addons that are using deadline as instance and representation collection should be done in host integrations (because they know what the output will be) -> this would also require some changes in ayon-core. The refactor would also probably resolve issues connected to review AOV filtering. But that is epic. |
I am also attaching here some info about SXR file format https://www.fxphd.com/fxblog/stereo-openexr-opensxr/. This is quite related to the post I was writing in discussion on our Discord thread.
|
I'm not sure user knows about it, he just sent renders to farm. Or I don't know if 3d artists do know what output it will create, I guess it is based on what renderer is used, and what is set in settings? In that case it is not that easy to resolve. But at the end, the plugin |
I have worked on that problem before. There are naming conventions originating from OpenEXR that can be used to support more than RGBA. |
What if we add an additional filter to the transcode settings that is like a regex or list of channels to transcode if present - if no matching channels, leave the file as is? |
Sure, but I think we should be able to handle single-channel files (A, Z or Y). |
Yup - which would then be what we'd add to the 'defaults' in the profiles? It does leave the question whether you really want to "transcode" a |
mmm... yes, if we want to keep things simple we should probably ignore data channels, like Z. how would you present the new transcode settings ?
|
Just wanted to flag this that came up on Discord here that mentioned using
AND
|
Good point @BigRoy. That would be even better. To be tested ! While I'm here, it looks like we could add deep image support quite easily: File: client/ayon_core/lib/transcoding.py
730: if input_info["attribs"].get("deep", False):
731: # flatten deep images: they are not natively supported by ffmpeg.
732: oiio_cmd.extend(["--flatten"])
733: |
Just adding @jakubjezek001's note that we'll also need to support stereo EXRs and may need to specify the "view" in the transcoding. @philippe-ynput for now - should we just add the channels choose in settings? Shall we do a quick call to discuss to get this one rolling? |
The approach now will be:
The new defaults would then include a channel mapping that would match the current behavior so it's backwards compatible, except for it not raising an error if the image does not contain any matching channel. We'd then just log a warning most likely. Aside of the linked OpenEXR page on channel names
I need to create a mapping in Settings where someone can state what channels in an image should map to which channel. Like being able to say:
And most likely even being able to support a regex (or fnmatch), like matching anything that endswith ayon-core/client/ayon_core/lib/transcoding.py Lines 320 to 403 in df95e25
The question then becomes: What's a decent |
Uh oh!
There was an error while loading. Please reload this page.
Is there an existing issue for this?
Current Behavior:
Over time we've had annoying reports where studios were unable to publish their renders because the publishing system (due to user configuration) tries to transcode e.g. a Z-depth AOV layer which may only have a named
Z
orY
channel and has noR
,G
,B
orA
channels and hence we don't really know what layer to process.Instead of this we could also implement detecting more cases, so that if
RGBA
is not found in the file that maybe we allow to fall back to channels:Here's metadata output for a file that fails without this PR:
It's an Arnold Z-depth render AOV.
Error logs
It's good to be aware that
ExtractReview
would also try to find a reviewable channel if it thinks it can't be processed by FFMPEG. Currently, ifExtractReview
can't find a suitable RGBA channel then it'd just skip generating the reviewable for it.However, ExtractOIIOTranscode runs before ExtractReview and would (or at least could?) usually perform the necessary conversion.
Expected Behavior:
Either:
OR
Version
1.0.0
What platform you are running on?
Windows
Steps To Reproduce:
Are there any labels you wish to add?
Relevant log output:
No response
Additional context:
We do have settings to "filter out" certain AOVs because we've had this issue before on
Extract Review
. The setting is atayon+settings://deadline/publish/ProcessSubmittedJobOnFarm/aov_filter
however it does not apply toExtractOIIOTranscode
because transcode may occur on something that isn't necessarily a reviewable (I suppose?)The text was updated successfully, but these errors were encountered: