StreamDiffusionTD 0.2.99🚀🖼️
Added 2025-08-21 16:12:05 +0000 UTCHey All !
I'm excited to have an update to StreamDiffusionTD. This update includes a built in cloud backend (select local or Daydream in the Install page) as well as better support for 5xxx and mac. A more exciting update is coming on the local side of things, but currently the cloud version has a few new tricks (multi-controlnet + tensorrt) that are definitely worth checking out in the meantime.
I am very excited to see a fresh wave of interest + innovation by Daydream and others in the open source community. Looking forward to continuing to see what you make with it.
Features // Updates:
- Daydream API supported ( non local - EARLY / LIMITED access >> link here )
- 5xxx support (this TOX only offers limited support - no tensorrt or v2v - coming shortly and almost ready for testing)
- Optimization boost across most gpu
- Better Mac installation
- Feedback loop Toggle on settings 1 page 👀
Grab the TOX here >> https://www.patreon.com/posts/137019143/
As always thank you for your support - I have more to share on StreamDiffusionTD, LOPs, and more soon - look out for more videos on my YouTube. I will be posting some better documentation and written guides as well on my website shortly. And please join the discord and say hello (make sure Discord is linked in connected apps on Patreon for role acess) !
Comments
in the file wrapper_td.py config_urls = { "sd15": "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/configs/stable-diffusion/v1-inference.yaml", "sd21": "https://raw.githubusercontent.com/Stability-AI/generative-models/main/configs/inference/sd_2_1.yaml", "sdxl": "https://raw.githubusercontent.com/Stability-AI/generative-models/main/configs/inference/sd_xl_base.yaml", "sdxl-turbo": "https://raw.githubusercontent.com/Stability-AI/generative-models/main/configs/inference/sd_xl_base.yaml", } we can see this, and when i copy and paste the link sd21 it appear break, ... but when we change the model the problem is the same very strange...
Bertrand Lepatre
2026-01-14 15:14:48 +0000 UTCHi, I’m using version 0.2.99 It works very well on both my Pc in the studio that uses rtx 3080 Ti and 4070. Im currently preparing a show to open next week and I’m getting this error on 2 brand new pc with rtx5060 : Exception : Failed to download config for sd21. I’ve ré-installed with cuda 12.8 Nothing has changed. Please help I’ve been struggling for 4 days now
David Gumbs 2
2026-01-14 14:14:02 +0000 UTCi tried cuda 11.8 python 3.10.9, always the same error, don't know, why it doesn't work for me. i m on win 11
Bertrand Lepatre
2026-01-14 03:58:24 +0000 UTCI'm getting this error after the first execution of StableDissusion: "======================================= Using TensorRT... ERROR: cannot import name 'cudart' from 'cuda' (unknown location)... ERROR: Failed to load TensorRT due to an unexpected error. This feature is Windows only ( I have Windows ) . Check the installation and restart the Stream." Anyone has the same problem? I have followed all the instructions about installation. I Also checked on HQImmersiveInteractive platform that i have but there is nothing about it I'm on python 3.11.9 and cuda 11.8
Mattia Minerva
2025-11-04 12:37:42 +0000 UTCsame problem, in multiple computers
Victor E Morales
2025-10-28 23:42:10 +0000 UTCSome problem...Daydream 400 in v0.2.99.. (Failed to create stream (HTTP 400)
Santa Fe
2025-10-28 23:10:53 +0000 UTChello there, first of all many thanks for you contribution to TD community, I'm a huge fan. any idea if is possible to run this using a AMD GPU?
Gabriel Worm
2025-10-23 02:33:04 +0000 UTCHave a trouble. streaming is active but in TD Server Connected, Not Streaming
MIkhail Loginev
2025-10-22 11:20:11 +0000 UTCVery curious about this too!
Robin - VR Learning Lab
2025-10-17 10:42:02 +0000 UTCHello! Will you support streamdiffusion V2?
Victor E Morales
2025-10-14 16:33:14 +0000 UTCHi, I'm trying to install this on mac and keep failing , any full tutorial for mac?
Alifka Zaidan
2025-10-09 21:06:28 +0000 UTCAwesome stuff, great work, thanks! I was wondering how to input negative prompts for example to prevent logos and text from showing.
Marcus Kitschke
2025-09-28 22:17:16 +0000 UTCThanks so much!
Luke Kantola
2025-09-28 03:37:51 +0000 UTCGot it working on both my RTX 5080 desktop and my 5070 Razer Blade 14. You need to install CUDA 12.8 for 5xxx GPUs. Works with a custom LoRA I made as well. Looking forward to the Tensor RT enabled version. Thank you!
Nefeli Manoudaki
2025-09-22 21:04:05 +0000 UTCCannot get it working on rtx 5090: it says RuntimeError: operator torchvision::nms does not exist
Salvo Nostrato
2025-09-18 15:17:30 +0000 UTCHey, new user here, just signed up and downloaded the .tox, clicked through to the installation guide on your website and the mouse animation gave me motion sickness within seconds. It's not something I'm usually sensitive to. Is there an easy way to switch it off that I missed? I found the gradient toggle but that doesn't help.
Dave Trevaskiss
2025-08-25 20:00:33 +0000 UTCHey Dot! thank you for everything you do for the TD community! Could you please point me in the right direction on how to troubleshoot this? "no module named 'torch.distributed.device_mesh" I have installend and updated everything, cuda, pip, pytorch etc. Is there a forum where i can discuss this? Thank you x
Guglielmo Padovan
2025-08-23 11:46:17 +0000 UTC
