Very cool! I've been learning Unreal Engine lately and PCG in particular, trying to build large procedural environments. I had the thought that exactly this functionality would be extremely useful but I didn't know where to start to implement it. Well done and I can't wait to see where this goes from here!
@EagleTopGaming4 күн бұрын
I like where this is going. Very cool.
@syntaxcg26 күн бұрын
hello any news to make this available? :)
@byronencar2 ай бұрын
for education, which one shouild i download?
@Proceduralit2 ай бұрын
There is no specific license for education in Orbolt. But both commercial and indie are free. I suggest you to try the commercial first.
@byronencar2 ай бұрын
@@Proceduralit Oh the reason i ask for education is because I'm currently a student and using the houdini education licenses issued out from our school.
@Proceduralit2 ай бұрын
@@byronencar Yeah, I got it but I don't have any information about the SideFX Education license, I think it's a commercial license. If your Houdini files are saved in .hip extension, so you need to download the commercial version of the tool. Or, maybe it's better to double-check it with SideFX support team.
@byronencar2 ай бұрын
@@Proceduralit Not sure what you mean by no info on SideFX education license. Do you mean Orbolt in general doesn't have education support? Still getting license errors with the .hip extension. I don't think as a student I'll be able to download the commercial version of the tool. Indie/apprentice will cause trouble for my files porting in back and forth between home setup and school workstation. Yeah, will have to check with the SideFX team or Orbolt support. Been exploring other avenues since this wasn't readily available. Appreciate the responses!
@Proceduralit2 ай бұрын
@@byronencar It seems the education license is like an Apprentice license, you can read more about it here: www.sidefx.com/community/student-access-to-houdini-education/ I suggest you contact the support team for your issue as the whole licensing system is handling with SideFX team and I don't have any control on it.
@piotrmichalak90314 ай бұрын
Outstanding!
@TheSlifki4 ай бұрын
thanks for the interesting tool, but all paths in the hou file are broken. this is a dislike. sorry man
@Proceduralit4 ай бұрын
You need to download the tile models from Kenney website. I mentioned it in the document and in the folder that you need to place the downloaded files.
@spectrewashere5 ай бұрын
Hi, it says there is an invalid license. I have done it through the Orbolt site authorization as well as download.
@Proceduralit5 ай бұрын
Hi, The licensing is completely handling by Sidefx, please contact their support for your issue. As far as I know, the licensing works better if you install it with orbolt as well.
@dj_multiple_one6 ай бұрын
hmm)) zenTangle) Cool thing, also thought about zentangle generator.
@user-mp3so9dx7x6 ай бұрын
how to make this shader in ue or unity
@Proceduralit6 ай бұрын
It's an UE project, you can download and open it in UE. Not sure about Unity but I think it's doable with the same algorithm.
@limdongyun39557 ай бұрын
That's amazing! Is it possible to get not only depth but also the Light Color Vercor?
@Proceduralit7 ай бұрын
It's possible but I think it's better in performance and visually to just export the light weight pass. Then adjust the weight and the color in UE.
@gravitomagneticpower8 ай бұрын
Thank you very much for this!!! I have explored the file. I don't get one thing. You have Right, Top, Front, Left, Bottom, Back, that is clear... Then you have MV: r+g/x+y and Alpha that is clear as well. But what is the blue channel in MV texture? Is it some ambient occlusion? Thank you!
@Proceduralit8 ай бұрын
It's something like the light scatter layer if I remember correctly.
@gravitomagneticpower8 ай бұрын
@@Proceduralit Thank you! :)
@derekcruickshank8 ай бұрын
*Promo sm* 😌
@newtimee10 ай бұрын
Good job, dude! But the only thing is that your sprite sheets have too much wasted space, you could use our texture space more efficiently
@Proceduralit10 ай бұрын
Thanks. Of course it's not a production ready asset. There're different solutions to improve it performance-wise based on the target device and platform.
@newtimee10 ай бұрын
@@Proceduralit anyway thank you for this! Now I’m trying to assemble the same shader on the UE5 with reacting pointlights and its unclear by now🥲
@Proceduralit10 ай бұрын
Good Luck @@newtimee
@milad_savar Жыл бұрын
Very Nice !!
@TaylorColpitts Жыл бұрын
Super cool!
@ahmadaljanadi9145 Жыл бұрын
sehr schon
@Displant Жыл бұрын
nice
@soudaarpe Жыл бұрын
Very good! Thanks for sharing!✌
@RMAFIRE Жыл бұрын
this is so weird, I love it ! great job 🚀🧨
@RMAFIRE Жыл бұрын
That’s creative!🙌✨
@SuperEvilstyle Жыл бұрын
awesome!
@RMAFIRE Жыл бұрын
That's great !! 💫💯✌
@RMAFIRE Жыл бұрын
Sensational !! 🤩💣
@syns369 Жыл бұрын
Is it available ?
@sebastianblatter7718 Жыл бұрын
Is this still working in 19.5?
@Proceduralit Жыл бұрын
It should but I haven't tested it.
@sebastianblatter7718 Жыл бұрын
@@Proceduralit Does not seem to work, sorry! Also tried the example files.
@Proceduralit Жыл бұрын
@@sebastianblatter7718 I'll check it out and update you. Thanks for informing.
@Proceduralit Жыл бұрын
I've tested it, it's working without any error. May I know what kind of error you get on your side?
@LoicWeil Жыл бұрын
Hi, When I press get sd models it returns "Failed to get sd models: String indices must be integers" Even after reading previous reviews, I don't understand where the problem is. I am using Houdini 19.5.493 and python 3.10.6 When I check the list of StableDiffusion checkpoints to make sure all checkpoints are installed, I only see "v1-5-pruned-emaonly.safetensors [6CE0161689]" in the list. THANKS
@RemitOtkin Жыл бұрын
Thanks a lot for your project! I've connected my project to Stable Diffusion and see all of the models of my AUTOMATIC1111, but for any reason, it doesn't work. I have a problem with Denied Access to the project folder, even with running as admin. Traceback (most recent call last): File "gen_SD_autoServer.generate", line 34, in <module> PermissionError: [Errno 13] Permission denied: 'C:/User....' Is it a common problem, or smth specific?
@Proceduralit Жыл бұрын
Unfortunately, I haven't faced with this error so far. May I know, which parameter of the tool is using the folder that needs permission? Is it the server installed folder, init_image, or output folder?
@RemitOtkin Жыл бұрын
@@Proceduralit I tried to set a folder like Desktop in both param (init img, render img) but the result was the same. Hope, will find the solution. Anyway, thanks)
@nicksushkevich Жыл бұрын
using version 2 with automation, get this error when trying to generate: Server Error: <Response [500]> - <bound method Response.json of <Response [500]>>
@Proceduralit Жыл бұрын
Based on the error message, your server is running and most likely there is an issue with the inputs. Are you sure you have set all the required parameters? github.com/proceduralit/StableDiffusion_Houdini/wiki/Setup-AUTOMATIC1111-Server kzread.info/dash/bejne/lZ2c07tqod27Zqg.html
@nicksushkevich Жыл бұрын
@@Proceduralit After double checking (thanks for your newest video) I managed to get it running. But when choosing Mantra, there are still error appears. Would it be possible in the future to generate directly from viewport without needing to render out input image?
@Proceduralit Жыл бұрын
@@nicksushkevich Thanks for the suggestion, it could be a handy feature. Right now, you can choose the openGL for rendering the init_image as it's very close to the viewport and the rendering time is close to zero.
@TraceMistDump Жыл бұрын
Does anyone meet this? AttributeError: 'module' object has no attribute 'auto_getSDModels' I'm using Houdini 19.5 with python 3.9 and still have no idea to fix this, because I'm a total newcomer
@Proceduralit Жыл бұрын
auto_getSDModels is a python function which returns the models installed on your local server. Please double-check if the server is running by loading the page in a browser.
@TraceMistDump Жыл бұрын
@@Proceduralit There's webui launching and I add --api in the arguments. But it's still show this message :( I don't know if I missing some thing in config the API. Because adding --api it's all I do
@Proceduralit Жыл бұрын
@@TraceMistDump Will try to reproduce this error on my side. Don't have any idea why it throws this error.
@OsmodinMedia Жыл бұрын
Thank you for this! I'm currently trying this and run into an error trying to render init_image for some reason. I can't post the whole error in comments but it ends with AttributeError NoneType object has no attribute prepareDirty
@Proceduralit Жыл бұрын
Thanks for reporting. May I know if you're using the latest version of Houdini? There is an issue with rendering in PDG networks that has been addressed with SideFX in the newer versions.
@theosoler8401 Жыл бұрын
Hi ! Thanks for your works ! I had no problem, until I entered the prompt and "generated" the image, I tried with the same geo as you and the same prompt but I did not have the same result. Is it necessary to have some basic texture on the geo?
@Proceduralit Жыл бұрын
Hi! I'm not very familiar with the Stable Diffusion algorithm but as far as I know, all the inputs are important and even a smaller difference in them will generate the different results. So, you should have all the parameters set with the same value and your input image should be exactly the same as mine. But, I suggest to ask someone who has better knowledge about the SD algorithm.
@roybatty2268 Жыл бұрын
Stable diffusion is a very interesting beast. Different hardware can affect the output image. A few extra commas, an innocuous setting difference or an update to Automatic1111 can affect the result etc. I have a ton of older images I can't come close to recreating with all the exact settings because of who knows for sure what reason.
@virgiliolorachaviano Жыл бұрын
great!
@Proceduralit Жыл бұрын
I forgot to mention the API mode should be enabled with the local server as it's not by default. github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/API
@vazelevz Жыл бұрын
Great work ! Already tried dreamstudio build .. but struggle with automatic1111 When I press get sd models it returns "Could not get SD-Models: string indices must be integers" Maybe you can tell what causes this ? I'm new to ai so definitely I've messed up something, but have no idea what )
@vimpassion Жыл бұрын
glad to know I'm not the only one getting that error.
@Proceduralit Жыл бұрын
@@vimpassion Hi guys! Sorry to hear you are having issue on server connection. 1- have you checked the server URL in a browser to make sure it's running? 2- Please double-check your URL path to make sure it's the same as the one with the browser. It seems the server is not running on your PC github.com/proceduralit/StableDiffusion_Houdini/wiki/Setup-AUTOMATIC1111-Server
@Proceduralit Жыл бұрын
I've checked the HDA with H19 & 19.5 with python 3x. But I think there are some issues with the python 2x version. May I ask which version of Houdini and Python is yours?
@vazelevz Жыл бұрын
@@Proceduralit houdini 19.5 and python 3.10.9 Maybe it's something with SD itself .. I'll try to reinstall later
@Proceduralit Жыл бұрын
@@vazelevz Thanks for the update. Yeah, I'll reinstall the server as well. I've installed it about 2 months ago, maybe something has been changed in the server API, maybe.
@vimpassion Жыл бұрын
When I click on GET SD Models button I get this error "Could not get SD-Models: string indices must be integers" Been trying to fix it for hours and cant figure out what's the problem. When I " print (response)" variable It outputs " response <Response [404]> "not sure if its a permission problems or the script cant find the path to the models\Stable-diffusion path. Could be that the callback script is not assigning kwargs["node"] if anyone has an idea what could be causing this error would be glad to hear it. I'm looking forward to using this tool locally.
@Proceduralit Жыл бұрын
Seems there is an issue with the server connection. May I know which version of Houdini and python you're trying the HDA with? Thanks.
@vimpassion Жыл бұрын
@@Proceduralit I'm using Houdini 19.5.435 - Py3.9 - also tried the latest Version 19.5.493
@Proceduralit Жыл бұрын
@@vimpassion Thanks for the update. I'll reinstall the server, hopefully I can reproduce the issue on my side.
@vimpassion Жыл бұрын
@@Proceduralit got it working now, just had to add set COMMANDLINE_ARGS=--api to the wabui-user.bat Thanks!
@notverydarkmagic Жыл бұрын
Dude this is awesome! Congrats!!
@Klonkel Жыл бұрын
Great tool! Is there any possibility of making this work with Karma XPU?
@Proceduralit Жыл бұрын
Hmmm, interesting idea. Will check it out. Thanks for the suggestion.
@NicoMahlerCG Жыл бұрын
omg i had to do a shot like 00:38 and didnt manage to get the displaced volume to stick to the animation. it just was a character moving through a world space volume noise. can somebody explain to me how he got that? incredible reel overall!
@Proceduralit Жыл бұрын
You can download it for free. www.orbolt.com/asset/Mohsen::mt_volumeDisplace::1.0
@NicoMahlerCG Жыл бұрын
@@Proceduralit oh damn. Thanks alot! :)
@dj_multiple_one Жыл бұрын
Absolutely amazing tool!
@jamesrussell6467 Жыл бұрын
Get an error about ModuleNotFoundError: No module named 'IPython'. odd as it is definitely installed.
@Proceduralit Жыл бұрын
Yeah, that's strange. May I know what is your Houdini version?
@jamesrussell6467 Жыл бұрын
@@Proceduralit H19.5.493 (Py3.9) Python 3.10.9 IPython 8.9.0 - but possibly a contributing factor. I am on Win11.
@Proceduralit Жыл бұрын
@@jamesrussell6467 Just for double-checking, the Houdini's package directory is not matter, IPython should be installed on the virtual environment's package directory(venv_dir\Lib\site-packages). github.com/proceduralit/StableDiffusion_Houdini/wiki/Setup-DreamStudio-Server
@jamesrussell6467 Жыл бұрын
@@Proceduralit interesting it looks like the batch file fails to create the content in the ds_venv folder. I will remove it all and go through the steps again.
@jamesrussell6467 Жыл бұрын
ok I think I sorted it. Python path I was feeding it was a shortcut linking to the python installation. Replaceing that with the correct path seems to be getting further. Think confusion arose from the way Win11 is working with Python.
@pojitonov Жыл бұрын
Looks like in the future instead of GPU rendering engines we will be using AI rendering engines.
@Proceduralit Жыл бұрын
Maybe
@SpinTurbo Жыл бұрын
Very nice work! Do you plan to release it? Does the z channel of the render feed into the diffusion process? or is it a midas Approximation?
@Proceduralit Жыл бұрын
Thanks. Yeah, I just need to do some minor adjustments and it will be released in my github. The depth data is calculated by stable diffusion on the fly. So Houdini has nothing to do with it.
@Proceduralit Жыл бұрын
It's been released! kzread.info/dash/bejne/lZ2c07tqod27Zqg.html
@AshkanZ Жыл бұрын
thats great man. how about mid journey?
@Proceduralit Жыл бұрын
Thanks, Ashkan 🙏 Yeah, having Midjourney in Houdini would be awesome but unfortunately, it doesn't provide the API for connecting, if I'm not mistaken.
@bondell1840 Жыл бұрын
WOW!
@KO28JOJO Жыл бұрын
Interesting! Would it work with Redshift in Hoduini ?
@Proceduralit Жыл бұрын
Most likely as I used the general mathematic nodes in the tool and I think all of them would be supported by Redshift. But I haven't tested it yet.
@KO28JOJO Жыл бұрын
@@Proceduralit thx. I tried it last. It did not work.
@Proceduralit Жыл бұрын
@@KO28JOJO Did you get any error about the unknown nodes?
@KO28JOJO Жыл бұрын
@@Proceduralit no, it just does not work. Maybe the math nodes are not supported in redshift
@mladenpajdic503 Жыл бұрын
@@Proceduralit I really liked to see it working in Redshift
@MaryCarrozzino Жыл бұрын
I would be really interested to watch how to build a network with functions similar to this in Unreal Engine!!
@Proceduralit Жыл бұрын
I think I still have the UE Material, will find and upload it on my github page. Will update you here soon, when it's ready.
@Proceduralit Жыл бұрын
You can find the UE material here: github.com/proceduralit/PI_Sandbox/tree/main/Content/GuidedNoise Implementing the whole system in UE from scratch is interesting and challenging.
@petewallace4376 Жыл бұрын
Yup! - got it working here! Had to add --user to the batch to get rid of some errors in pip install
@petewallace4376 Жыл бұрын
Although the COPS bit isn't working for me. Local variable 'init_image' not found. Local variable 'pdg_output' not found.
@Proceduralit Жыл бұрын
@@petewallace4376 I think you haven't defined the output folders for the init_image and the stable diffusion images.
@petewallace4376 Жыл бұрын
@@Proceduralit Those are set to $HIP/folders and seem to be working in terms of init and out being created - but perhaps it isn't making it into the COPs part of the process? Also, I'm setting the init to 1024 --- but the output is still 512?
@Proceduralit Жыл бұрын
@@petewallace4376 It's strange. Will double-check it later. Have you tried the Refresh Output Viewer button? You also can check the details by double or middle click on the work_item. Could you please check if you have the init_image and out_image with the correct address there?
@petewallace4376 Жыл бұрын
@@Proceduralit yup - but that stays 1x1 pixel. The copnet shows errors for the two incoming file nodes.
@petewallace4376 Жыл бұрын
I've been using a Colab for this, but would love to give your HDA a go!
@Proceduralit Жыл бұрын
Thanks Pete. Right now working on it to supports a public server. (Most likely, DreamStudio) Will release it soon.
@petewallace4376 Жыл бұрын
@@Proceduralit That's great! Local SD installs will be great for hammering iterations or animated sequences overnight also. I'm hoping to stumble across a TOP network that does ML upscaling also. But i realise it's a shifting landscape.
Пікірлер
Very cool! I've been learning Unreal Engine lately and PCG in particular, trying to build large procedural environments. I had the thought that exactly this functionality would be extremely useful but I didn't know where to start to implement it. Well done and I can't wait to see where this goes from here!
I like where this is going. Very cool.
hello any news to make this available? :)
for education, which one shouild i download?
There is no specific license for education in Orbolt. But both commercial and indie are free. I suggest you to try the commercial first.
@@Proceduralit Oh the reason i ask for education is because I'm currently a student and using the houdini education licenses issued out from our school.
@@byronencar Yeah, I got it but I don't have any information about the SideFX Education license, I think it's a commercial license. If your Houdini files are saved in .hip extension, so you need to download the commercial version of the tool. Or, maybe it's better to double-check it with SideFX support team.
@@Proceduralit Not sure what you mean by no info on SideFX education license. Do you mean Orbolt in general doesn't have education support? Still getting license errors with the .hip extension. I don't think as a student I'll be able to download the commercial version of the tool. Indie/apprentice will cause trouble for my files porting in back and forth between home setup and school workstation. Yeah, will have to check with the SideFX team or Orbolt support. Been exploring other avenues since this wasn't readily available. Appreciate the responses!
@@byronencar It seems the education license is like an Apprentice license, you can read more about it here: www.sidefx.com/community/student-access-to-houdini-education/ I suggest you contact the support team for your issue as the whole licensing system is handling with SideFX team and I don't have any control on it.
Outstanding!
thanks for the interesting tool, but all paths in the hou file are broken. this is a dislike. sorry man
You need to download the tile models from Kenney website. I mentioned it in the document and in the folder that you need to place the downloaded files.
Hi, it says there is an invalid license. I have done it through the Orbolt site authorization as well as download.
Hi, The licensing is completely handling by Sidefx, please contact their support for your issue. As far as I know, the licensing works better if you install it with orbolt as well.
hmm)) zenTangle) Cool thing, also thought about zentangle generator.
how to make this shader in ue or unity
It's an UE project, you can download and open it in UE. Not sure about Unity but I think it's doable with the same algorithm.
That's amazing! Is it possible to get not only depth but also the Light Color Vercor?
It's possible but I think it's better in performance and visually to just export the light weight pass. Then adjust the weight and the color in UE.
Thank you very much for this!!! I have explored the file. I don't get one thing. You have Right, Top, Front, Left, Bottom, Back, that is clear... Then you have MV: r+g/x+y and Alpha that is clear as well. But what is the blue channel in MV texture? Is it some ambient occlusion? Thank you!
It's something like the light scatter layer if I remember correctly.
@@Proceduralit Thank you! :)
*Promo sm* 😌
Good job, dude! But the only thing is that your sprite sheets have too much wasted space, you could use our texture space more efficiently
Thanks. Of course it's not a production ready asset. There're different solutions to improve it performance-wise based on the target device and platform.
@@Proceduralit anyway thank you for this! Now I’m trying to assemble the same shader on the UE5 with reacting pointlights and its unclear by now🥲
Good Luck @@newtimee
Very Nice !!
Super cool!
sehr schon
nice
Very good! Thanks for sharing!✌
this is so weird, I love it ! great job 🚀🧨
That’s creative!🙌✨
awesome!
That's great !! 💫💯✌
Sensational !! 🤩💣
Is it available ?
Is this still working in 19.5?
It should but I haven't tested it.
@@Proceduralit Does not seem to work, sorry! Also tried the example files.
@@sebastianblatter7718 I'll check it out and update you. Thanks for informing.
I've tested it, it's working without any error. May I know what kind of error you get on your side?
Hi, When I press get sd models it returns "Failed to get sd models: String indices must be integers" Even after reading previous reviews, I don't understand where the problem is. I am using Houdini 19.5.493 and python 3.10.6 When I check the list of StableDiffusion checkpoints to make sure all checkpoints are installed, I only see "v1-5-pruned-emaonly.safetensors [6CE0161689]" in the list. THANKS
Thanks a lot for your project! I've connected my project to Stable Diffusion and see all of the models of my AUTOMATIC1111, but for any reason, it doesn't work. I have a problem with Denied Access to the project folder, even with running as admin. Traceback (most recent call last): File "gen_SD_autoServer.generate", line 34, in <module> PermissionError: [Errno 13] Permission denied: 'C:/User....' Is it a common problem, or smth specific?
Unfortunately, I haven't faced with this error so far. May I know, which parameter of the tool is using the folder that needs permission? Is it the server installed folder, init_image, or output folder?
@@Proceduralit I tried to set a folder like Desktop in both param (init img, render img) but the result was the same. Hope, will find the solution. Anyway, thanks)
using version 2 with automation, get this error when trying to generate: Server Error: <Response [500]> - <bound method Response.json of <Response [500]>>
Based on the error message, your server is running and most likely there is an issue with the inputs. Are you sure you have set all the required parameters? github.com/proceduralit/StableDiffusion_Houdini/wiki/Setup-AUTOMATIC1111-Server kzread.info/dash/bejne/lZ2c07tqod27Zqg.html
@@Proceduralit After double checking (thanks for your newest video) I managed to get it running. But when choosing Mantra, there are still error appears. Would it be possible in the future to generate directly from viewport without needing to render out input image?
@@nicksushkevich Thanks for the suggestion, it could be a handy feature. Right now, you can choose the openGL for rendering the init_image as it's very close to the viewport and the rendering time is close to zero.
Does anyone meet this? AttributeError: 'module' object has no attribute 'auto_getSDModels' I'm using Houdini 19.5 with python 3.9 and still have no idea to fix this, because I'm a total newcomer
auto_getSDModels is a python function which returns the models installed on your local server. Please double-check if the server is running by loading the page in a browser.
@@Proceduralit There's webui launching and I add --api in the arguments. But it's still show this message :( I don't know if I missing some thing in config the API. Because adding --api it's all I do
@@TraceMistDump Will try to reproduce this error on my side. Don't have any idea why it throws this error.
Thank you for this! I'm currently trying this and run into an error trying to render init_image for some reason. I can't post the whole error in comments but it ends with AttributeError NoneType object has no attribute prepareDirty
Thanks for reporting. May I know if you're using the latest version of Houdini? There is an issue with rendering in PDG networks that has been addressed with SideFX in the newer versions.
Hi ! Thanks for your works ! I had no problem, until I entered the prompt and "generated" the image, I tried with the same geo as you and the same prompt but I did not have the same result. Is it necessary to have some basic texture on the geo?
Hi! I'm not very familiar with the Stable Diffusion algorithm but as far as I know, all the inputs are important and even a smaller difference in them will generate the different results. So, you should have all the parameters set with the same value and your input image should be exactly the same as mine. But, I suggest to ask someone who has better knowledge about the SD algorithm.
Stable diffusion is a very interesting beast. Different hardware can affect the output image. A few extra commas, an innocuous setting difference or an update to Automatic1111 can affect the result etc. I have a ton of older images I can't come close to recreating with all the exact settings because of who knows for sure what reason.
great!
I forgot to mention the API mode should be enabled with the local server as it's not by default. github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/API
Great work ! Already tried dreamstudio build .. but struggle with automatic1111 When I press get sd models it returns "Could not get SD-Models: string indices must be integers" Maybe you can tell what causes this ? I'm new to ai so definitely I've messed up something, but have no idea what )
glad to know I'm not the only one getting that error.
@@vimpassion Hi guys! Sorry to hear you are having issue on server connection. 1- have you checked the server URL in a browser to make sure it's running? 2- Please double-check your URL path to make sure it's the same as the one with the browser. It seems the server is not running on your PC github.com/proceduralit/StableDiffusion_Houdini/wiki/Setup-AUTOMATIC1111-Server
I've checked the HDA with H19 & 19.5 with python 3x. But I think there are some issues with the python 2x version. May I ask which version of Houdini and Python is yours?
@@Proceduralit houdini 19.5 and python 3.10.9 Maybe it's something with SD itself .. I'll try to reinstall later
@@vazelevz Thanks for the update. Yeah, I'll reinstall the server as well. I've installed it about 2 months ago, maybe something has been changed in the server API, maybe.
When I click on GET SD Models button I get this error "Could not get SD-Models: string indices must be integers" Been trying to fix it for hours and cant figure out what's the problem. When I " print (response)" variable It outputs " response <Response [404]> "not sure if its a permission problems or the script cant find the path to the models\Stable-diffusion path. Could be that the callback script is not assigning kwargs["node"] if anyone has an idea what could be causing this error would be glad to hear it. I'm looking forward to using this tool locally.
Seems there is an issue with the server connection. May I know which version of Houdini and python you're trying the HDA with? Thanks.
@@Proceduralit I'm using Houdini 19.5.435 - Py3.9 - also tried the latest Version 19.5.493
@@vimpassion Thanks for the update. I'll reinstall the server, hopefully I can reproduce the issue on my side.
@@Proceduralit got it working now, just had to add set COMMANDLINE_ARGS=--api to the wabui-user.bat Thanks!
Dude this is awesome! Congrats!!
Great tool! Is there any possibility of making this work with Karma XPU?
Hmmm, interesting idea. Will check it out. Thanks for the suggestion.
omg i had to do a shot like 00:38 and didnt manage to get the displaced volume to stick to the animation. it just was a character moving through a world space volume noise. can somebody explain to me how he got that? incredible reel overall!
You can download it for free. www.orbolt.com/asset/Mohsen::mt_volumeDisplace::1.0
@@Proceduralit oh damn. Thanks alot! :)
Absolutely amazing tool!
Get an error about ModuleNotFoundError: No module named 'IPython'. odd as it is definitely installed.
Yeah, that's strange. May I know what is your Houdini version?
@@Proceduralit H19.5.493 (Py3.9) Python 3.10.9 IPython 8.9.0 - but possibly a contributing factor. I am on Win11.
@@jamesrussell6467 Just for double-checking, the Houdini's package directory is not matter, IPython should be installed on the virtual environment's package directory(venv_dir\Lib\site-packages). github.com/proceduralit/StableDiffusion_Houdini/wiki/Setup-DreamStudio-Server
@@Proceduralit interesting it looks like the batch file fails to create the content in the ds_venv folder. I will remove it all and go through the steps again.
ok I think I sorted it. Python path I was feeding it was a shortcut linking to the python installation. Replaceing that with the correct path seems to be getting further. Think confusion arose from the way Win11 is working with Python.
Looks like in the future instead of GPU rendering engines we will be using AI rendering engines.
Maybe
Very nice work! Do you plan to release it? Does the z channel of the render feed into the diffusion process? or is it a midas Approximation?
Thanks. Yeah, I just need to do some minor adjustments and it will be released in my github. The depth data is calculated by stable diffusion on the fly. So Houdini has nothing to do with it.
It's been released! kzread.info/dash/bejne/lZ2c07tqod27Zqg.html
thats great man. how about mid journey?
Thanks, Ashkan 🙏 Yeah, having Midjourney in Houdini would be awesome but unfortunately, it doesn't provide the API for connecting, if I'm not mistaken.
WOW!
Interesting! Would it work with Redshift in Hoduini ?
Most likely as I used the general mathematic nodes in the tool and I think all of them would be supported by Redshift. But I haven't tested it yet.
@@Proceduralit thx. I tried it last. It did not work.
@@KO28JOJO Did you get any error about the unknown nodes?
@@Proceduralit no, it just does not work. Maybe the math nodes are not supported in redshift
@@Proceduralit I really liked to see it working in Redshift
I would be really interested to watch how to build a network with functions similar to this in Unreal Engine!!
I think I still have the UE Material, will find and upload it on my github page. Will update you here soon, when it's ready.
You can find the UE material here: github.com/proceduralit/PI_Sandbox/tree/main/Content/GuidedNoise Implementing the whole system in UE from scratch is interesting and challenging.
Yup! - got it working here! Had to add --user to the batch to get rid of some errors in pip install
Although the COPS bit isn't working for me. Local variable 'init_image' not found. Local variable 'pdg_output' not found.
@@petewallace4376 I think you haven't defined the output folders for the init_image and the stable diffusion images.
@@Proceduralit Those are set to $HIP/folders and seem to be working in terms of init and out being created - but perhaps it isn't making it into the COPs part of the process? Also, I'm setting the init to 1024 --- but the output is still 512?
@@petewallace4376 It's strange. Will double-check it later. Have you tried the Refresh Output Viewer button? You also can check the details by double or middle click on the work_item. Could you please check if you have the init_image and out_image with the correct address there?
@@Proceduralit yup - but that stays 1x1 pixel. The copnet shows errors for the two incoming file nodes.
I've been using a Colab for this, but would love to give your HDA a go!
Thanks Pete. Right now working on it to supports a public server. (Most likely, DreamStudio) Will release it soon.
@@Proceduralit That's great! Local SD installs will be great for hammering iterations or animated sequences overnight also. I'm hoping to stumble across a TOP network that does ML upscaling also. But i realise it's a shifting landscape.
Looking forward to whatever you can share.
@@roybatty2268 Very soon, it's almost ready.
Awesome!.