Hey,
I've been trying higher viewport quality settings and for the most part things are looking great. I mainly seem to get odd results switching better the transparency options for things like glass.
It seems going to medium / high quality does some things like losing the reflection or weirdly showing through opaque objects like a wine bottle label for example
Just wondering if theres any knowledge about this? i have a Nvidia GTX 1080 card
Also seems similar to this question i think?:
https://www.sidefx.com/forum/topic/74455/ [www.sidefx.com]
Many Thanks,
~Craig
Found 30 posts.
Search results Show results as topic list.
Technical Discussion » Viewport Transparency Behaviour
- mrtonks
- 54 posts
- Offline
Technical Discussion » Performance Monitor / Python Workflows
- mrtonks
- 54 posts
- Offline
Thanks man! Appreciate the reply!
Should of followed up in here; i did talk to sidefx, confirmed the limitations and get this logged as an RFE for extra python functionality so we’ll see if anything comes of that
Should of followed up in here; i did talk to sidefx, confirmed the limitations and get this logged as an RFE for extra python functionality so we’ll see if anything comes of that
Technical Discussion » Performance Monitor / Python Workflows
- mrtonks
- 54 posts
- Offline
Hey All,
I was dabbling with chucking out a per frame performance monitor file whilst i cache (start, cook geo, stop, save profile). Largely seems to work as expected but theres a few things i'm unsure about:
1. The performance monitor seems to be getting clogged up with all these profiles, i cant see how to purge them once saved? it sounds likewould do it but it doesnt look like you can both stop / save the stats and cancel?
2. I wanted to do this accross pre / post frame scripts but i couldnt find how you would reference the profile again if you dont have the variable set in one script like the helps example (). I thought there would be a kind of way of doing this? I also tried active profile but it didnt seem to be quite right
Anyone done workflows with this and got useful tips?
~Craig
I was dabbling with chucking out a per frame performance monitor file whilst i cache (start, cook geo, stop, save profile). Largely seems to work as expected but theres a few things i'm unsure about:
1. The performance monitor seems to be getting clogged up with all these profiles, i cant see how to purge them once saved? it sounds like
cancel()
2. I wanted to do this accross pre / post frame scripts but i couldnt find how you would reference the profile again if you dont have the variable set in one script like the helps example (
profile = hou.perfMon.startProfile()
hou.perfMon.Profile(profilename).stop()
Anyone done workflows with this and got useful tips?
~Craig
Technical Discussion » Hqueue - Understanding Child Jobs order
- mrtonks
- 54 posts
- Offline
Hey,
I'm trying to set off file caches that are quite heavy and as such looking at splitting it into jobs. However i cant figure how to make sure it respects order and dependency?
If i have 2 streams of geo rops each ending in a hqueue sim node each, plugged into a merge then a last hqueue sim node i get the correct looking network on Hqueue:
Parent
> Child1
> Child2
However its running to process both the children at the same time and failing.
I also tried with everything in a single tree but that seems very suspect too. Is there something i'm missing? Or a better way of Managing this?
Thanks!
~Craig
I'm trying to set off file caches that are quite heavy and as such looking at splitting it into jobs. However i cant figure how to make sure it respects order and dependency?
If i have 2 streams of geo rops each ending in a hqueue sim node each, plugged into a merge then a last hqueue sim node i get the correct looking network on Hqueue:
Parent
> Child1
> Child2
However its running to process both the children at the same time and failing.
I also tried with everything in a single tree but that seems very suspect too. Is there something i'm missing? Or a better way of Managing this?
Thanks!
~Craig
Technical Discussion » HQueue and more detailed progress log?
- mrtonks
- 54 posts
- Offline
Hey,
Before i start fiddling too much - has anyone done much work with Hqueue processing lots of geometry rops? I'm looking for the best way to get verbose progress through the log and cant see anything out of the box to do this?
I guess ideal would be something like using the pre / post frame scripts to spit
Node - Frame
into the log for example. Would that be the best approach / is it more complicated?
Edit: i just noticed some python code i was spitting out for debugging at sop level is going in anyway so i can even do that!
Thanks!
~Craig
Before i start fiddling too much - has anyone done much work with Hqueue processing lots of geometry rops? I'm looking for the best way to get verbose progress through the log and cant see anything out of the box to do this?
I guess ideal would be something like using the pre / post frame scripts to spit
Node - Frame
into the log for example. Would that be the best approach / is it more complicated?
Edit: i just noticed some python code i was spitting out for debugging at sop level is going in anyway so i can even do that!
Thanks!
~Craig
Edited by mrtonks - June 3, 2019 06:28:31
PDG/TOPs » Filtering highest file / folder version
- mrtonks
- 54 posts
- Offline
This is a great start Chris! So what i'm trying to do is that, butttt combined with partitioning by shot.
I've re-attached the file and added in a basic example of the work items having a “shot” string attribute - so we've hypothetically found all versions in these shots and want to chuck out all but the highest
I've re-attached the file and added in a basic example of the work items having a “shot” string attribute - so we've hypothetically found all versions in these shots and want to chuck out all but the highest
PDG/TOPs » Render all OUT context rops in a file
- mrtonks
- 54 posts
- Offline
I'll try and recreate the test i was doing at work at home! firewalls and all that. making a hash of explaining it on here haha
Maybe simpler way of explaining what i'm after / found:
I don't feel like the rop fetch is quite geared towards daisy chains? but no other TOPS seem obviously geared to this either - i say this fully hoping i'm being very stupid and missing something!
Maybe simpler way of explaining what i'm after / found:
- can the rop fetch look at any number of Hips (it can if passing file as an attribute)
- It specifies a rop, but can it evaluate all the chained ROPS plugged into the one specified (it does)
- How do strict frame ranges, number of rops and the rop fetch frame range determine work items generated (it seems it just makes a work item per rop fetch frame range per file, and booleans any frame ranges within)
I don't feel like the rop fetch is quite geared towards daisy chains? but no other TOPS seem obviously geared to this either - i say this fully hoping i'm being very stupid and missing something!
PDG/TOPs » Render all OUT context rops in a file
- mrtonks
- 54 posts
- Offline
Had a bit of time to return to this, i seemed to reliably get some errored work items each time but they do seem to be blips from what i can tell? it only says:
Error: The Attempted Operation Failed
The Rop fetch does largely seemed to be working right though i need to check with a more complex structure.
I am a bit cloudy on the work item generation though! So the ROP fetch hip path takes @file (my csv list of 3 file paths) and is set to frames 1-240. This makes 720 work items understandably - however one file for example has caches with strict ranges of 1-1 and another of 1-12. The correct ranges are output but i'm not sure what all those work items do if that scenes caching around 13 frames? I guess i'm almost expecting it to make one item per frame of cache?
Error: The Attempted Operation Failed
The Rop fetch does largely seemed to be working right though i need to check with a more complex structure.
I am a bit cloudy on the work item generation though! So the ROP fetch hip path takes @file (my csv list of 3 file paths) and is set to frames 1-240. This makes 720 work items understandably - however one file for example has caches with strict ranges of 1-1 and another of 1-12. The correct ranges are output but i'm not sure what all those work items do if that scenes caching around 13 frames? I guess i'm almost expecting it to make one item per frame of cache?
PDG/TOPs » Filtering highest file / folder version
- mrtonks
- 54 posts
- Offline
Technical Discussion » Vellum - Consistent Frame for Hair
- mrtonks
- 54 posts
- Offline
Hey,
Little confused over vellums handling of a hairs frame for matching pre / post simulation and just wanted to clarify for sanity!
As an excercise i have Rest hairs and Sim hairs i want to match some sort of orient / matrix on:
1. To start the Groom is guide deformed and creating Orient - which appears to be per hair
Normal, Tangent and Bitangents made with polyframe
These are run for both the rest and simulation hairs
2. Passing the sim hairs through Vellum Hair Constraints the orient changes to per point, regardless of having compute missing orientation on or off
3. Coming out of the sim the polyframe attributes dont appear to update. Recreating polyframe after results in some dodgy spinning, but orient is doing exactly as it should. Though it no longer matches the rest hairs.
I know vellums clever with this stuff internally, so what should i be doing to see vectors / matrices come through accordingly deformed? or would i need to run the rest hairs through a vellum hair constraint node to match up and rely on the vellum orientation attribute?
Little confused over vellums handling of a hairs frame for matching pre / post simulation and just wanted to clarify for sanity!
As an excercise i have Rest hairs and Sim hairs i want to match some sort of orient / matrix on:
1. To start the Groom is guide deformed and creating Orient - which appears to be per hair
Normal, Tangent and Bitangents made with polyframe
These are run for both the rest and simulation hairs
2. Passing the sim hairs through Vellum Hair Constraints the orient changes to per point, regardless of having compute missing orientation on or off
3. Coming out of the sim the polyframe attributes dont appear to update. Recreating polyframe after results in some dodgy spinning, but orient is doing exactly as it should. Though it no longer matches the rest hairs.
I know vellums clever with this stuff internally, so what should i be doing to see vectors / matrices come through accordingly deformed? or would i need to run the rest hairs through a vellum hair constraint node to match up and rely on the vellum orientation attribute?
PDG/TOPs » Filtering highest file / folder version
- mrtonks
- 54 posts
- Offline
Thanks Ken!
Theres two problems with that though:
1. I wont always know the largest index, so i need some attribute promote style way of finding the max. Which is possible for all work items at once, but…
2. the work items belong to different folders. So i need the threshold for each group / partition of items
Essentially looking at Partition by folder attribute, loop over all items in that group and get the max - then do filter by expression. I just havnt got my head around the python code to do it and pass everything back out correctly
Attribute promote on work items or by partitioned items would be a great addition in the future!
Theres two problems with that though:
1. I wont always know the largest index, so i need some attribute promote style way of finding the max. Which is possible for all work items at once, but…
2. the work items belong to different folders. So i need the threshold for each group / partition of items
Essentially looking at Partition by folder attribute, loop over all items in that group and get the max - then do filter by expression. I just havnt got my head around the python code to do it and pass everything back out correctly
Attribute promote on work items or by partitioned items would be a great addition in the future!
PDG/TOPs » Filtering highest file / folder version
- mrtonks
- 54 posts
- Offline
Hey,
Painfully close on this but wondering what the intended workflow would be for processing highest values only.
As an example, i'm using a file pattern glob so that given a number of shots i can get all the hip files per shot - that gives me lots of useful info including an index based on file versions!
I'd like to then go forward with just the highest index. If it was geo i'd do an attrib promote to get the max and filter anything less - would the same logic but with a python script make most sense? or is there a more elegant solution i'm missing?
~Craig
Painfully close on this but wondering what the intended workflow would be for processing highest values only.
As an example, i'm using a file pattern glob so that given a number of shots i can get all the hip files per shot - that gives me lots of useful info including an index based on file versions!
I'd like to then go forward with just the highest index. If it was geo i'd do an attrib promote to get the max and filter anything less - would the same logic but with a python script make most sense? or is there a more elegant solution i'm missing?
~Craig
PDG/TOPs » Render all OUT context rops in a file
- mrtonks
- 54 posts
- Offline
Hey,
Just proof of concepting automatically running caches. my test looks like:
> Three shots with 2 geometry rops daisy chained
> first has a strict one frame cache, the other has a longer frame range
The two main options seem to be:
ROP Fetch from external file - it sort of does what i want, but i'm getting errors i think from it not properly working with the strict frame range and dependencies. I want it to move onto more complex out networks after.
Server Session - attempting to use python to file load then hit the render button, but hitting syntax problems passing a file path through. And for a more complex scene i'm not sure if it will calculate node by node as it seems to prefer frame by frame in normal sessions (have posted in technical about this)
Am i running down the right route?
Just proof of concepting automatically running caches. my test looks like:
> Three shots with 2 geometry rops daisy chained
> first has a strict one frame cache, the other has a longer frame range
The two main options seem to be:
ROP Fetch from external file - it sort of does what i want, but i'm getting errors i think from it not properly working with the strict frame range and dependencies. I want it to move onto more complex out networks after.
Server Session - attempting to use python to file load then hit the render button, but hitting syntax problems passing a file path through. And for a more complex scene i'm not sure if it will calculate node by node as it seems to prefer frame by frame in normal sessions (have posted in technical about this)
Am i running down the right route?
Technical Discussion » OUT Rendering Dependencies?
- mrtonks
- 54 posts
- Offline
Hey,
I'm out of a pipeline for the first time in a while and struggling with rendering dependencies. I thought the default was Node By Node but going to my final node and hitting render is making the network calculate frame by frame (unless i go to Controls.. and choose Node By Node)
Is it only hython / batch that goes node by node at its default? Whats the easiest way of forcing node by node by default?
~Craig
I'm out of a pipeline for the first time in a while and struggling with rendering dependencies. I thought the default was Node By Node but going to my final node and hitting render is making the network calculate frame by frame (unless i go to Controls.. and choose Node By Node)
Is it only hython / batch that goes node by node at its default? Whats the easiest way of forcing node by node by default?
~Craig
Technical Discussion » Set Default Wiring Style to rounded?
- mrtonks
- 54 posts
- Offline
Thats fantastic - its only a small thing but would be great as an option.
Thanks!
Thanks!
mtucker
No, there's no way to set the default. The wiring style is saved with the network now (since any network only looks good when viewed with the wiring style used to build it). But I didn't think to provide a way to pick a default wiring style for new networks. There is already an RFE submitted for this.
Technical Discussion » Set Default Wiring Style to rounded?
- mrtonks
- 54 posts
- Offline
This may be * really* simple… but i cant for the life of me find where to set the wiring style to “Rounded” by default?
Houdini Indie and Apprentice » Cloud Properties Python Error
- mrtonks
- 54 posts
- Offline
i'm running houdini indie 15.5.480.
I'm trying to set up cloud rendering, and after filling out the info in Cloud Properties i get this error pop up everytime:
Traceback (most recent call last):
File “CPROGRA~1/SIDEEF~1/HOUDIN~1.480/houdini/python2.7libs\cloud.py”, line 641, in <lambda>
lambda:_onCloseSetUpCloudDialog(dialogs))
File “CPROGRA~1/SIDEEF~1/HOUDIN~1.480/houdini/python2.7libs\cloud.py”, line 680, in _onCloseSetUpCloudDialog
_passphraseEncryptKey(secret_access_key, passphrase))
File “CPROGRA~1/SIDEEF~1/HOUDIN~1.480/houdini/python2.7libs\cloud.py”, line 392, in _passphraseEncryptKey
Crypto.Hash.SHA, salt, passphrase, len(data_to_encrypt))
File “C:\PROGRA~1\SIDEEF~1\HOUDIN~1.480\python27\lib\site-packages\paramiko\util.py”, line 157, in generate_key_bytes
hash_obj = hash_alg()
TypeError: ‘module’ object is not callable
Seems to be the same as this post below, can anyone shed some light on what causes this?
https://www.sidefx.com/forum/topic/43471/ [sidefx.com]
I'm trying to set up cloud rendering, and after filling out the info in Cloud Properties i get this error pop up everytime:
Traceback (most recent call last):
File “CPROGRA~1/SIDEEF~1/HOUDIN~1.480/houdini/python2.7libs\cloud.py”, line 641, in <lambda>
lambda:_onCloseSetUpCloudDialog(dialogs))
File “CPROGRA~1/SIDEEF~1/HOUDIN~1.480/houdini/python2.7libs\cloud.py”, line 680, in _onCloseSetUpCloudDialog
_passphraseEncryptKey(secret_access_key, passphrase))
File “CPROGRA~1/SIDEEF~1/HOUDIN~1.480/houdini/python2.7libs\cloud.py”, line 392, in _passphraseEncryptKey
Crypto.Hash.SHA, salt, passphrase, len(data_to_encrypt))
File “C:\PROGRA~1\SIDEEF~1\HOUDIN~1.480\python27\lib\site-packages\paramiko\util.py”, line 157, in generate_key_bytes
hash_obj = hash_alg()
TypeError: ‘module’ object is not callable
Seems to be the same as this post below, can anyone shed some light on what causes this?
https://www.sidefx.com/forum/topic/43471/ [sidefx.com]
Technical Discussion » Missing Cloth Features?
- mrtonks
- 54 posts
- Offline
Hi,
I must be missing something obvious here, but i seem to not have a lot of cloth related features:
According to the helpcard i'm missing from the cloth shelf: Delete Constraint, Paint Tearable and Attach constraint to animation.
Also according to helpcard to set up tearing there should be a Capabilities tab on the cloth solver, a Cloth Tearing Properties dynamics node… possibly more missing than i know.
This was with 12.5.316.22 and 12.5.371 under Linux and 12.5.371 under Windows 8.
Anyone else have the same?
I must be missing something obvious here, but i seem to not have a lot of cloth related features:
According to the helpcard i'm missing from the cloth shelf: Delete Constraint, Paint Tearable and Attach constraint to animation.
Also according to helpcard to set up tearing there should be a Capabilities tab on the cloth solver, a Cloth Tearing Properties dynamics node… possibly more missing than i know.
This was with 12.5.316.22 and 12.5.371 under Linux and 12.5.371 under Windows 8.
Anyone else have the same?
Technical Discussion » 12.5 Centroid Variables Issue?
- mrtonks
- 54 posts
- Offline
GCX, GCY, GCY indeed all work fine
And as mentioned, it appears $CE* variables do work internally - but don't feedback the correct values or pass the correct values to the viewport handle.
I have submitted this as a bug
And as mentioned, it appears $CE* variables do work internally - but don't feedback the correct values or pass the correct values to the viewport handle.
I have submitted this as a bug
Technical Discussion » 12.5 Centroid Variables Issue?
- mrtonks
- 54 posts
- Offline
Someone else then, must be a bug?
Have attached a simple example, using $CE* for the pivot and -$CE* for the transform to return the sphere to center.
Weird thing is, i am getting the right result for the centroid variables version - just the values shown are all 0, compared to centroid() that is giving me correct values.
Have attached a simple example, using $CE* for the pivot and -$CE* for the transform to return the sphere to center.
Weird thing is, i am getting the right result for the centroid variables version - just the values shown are all 0, compared to centroid() that is giving me correct values.
-
- Quick Links