This is a heavily modified workflow from LTX-2.3 All-In-One workflow for RTX 3060 with 12 GB VRAM + 32 GB RAM - v1.0 | LTX Video Workflows | Civitai.
All credits go to the creator for his excellent workflow.
I modified it using the settings and models from the official 10Eros workflow TenStrip/LTX2.3-10Eros_Workflows at main
Plus I added the "Prompt Relay Encode (Timeline)" and some other Q.O.L. settings.
All instructions and downloads are in the workflow.
Description
This is a huge update, better quality, following the recent changes from the official 10Eros workflow. New Distilled LoRA, download links inside the workflow. Better and more control for First - Last Frame and First - Middle - Last Frame, with the blue selector nodes, Please read carefully the instructions in the blue note node. Do not mess with the black selector nodes, unless you want faster generation for testing purposing. READ all the notes in the workflow, they contain useful info and instructions.
FAQ
Comments (12)
This is looking very promising thanks!
Is any chance you could add a PREVIEW node, so I can preview how my prompt is generating while it works? Or is it easy enough to add an LTS preview node to your flow?
Hi, the lowest black selector node (above "Clip length in seconds" node), have a "enable sampling preview" turn it to "Yes", the preview is shown in the "processing" node. Regards
Thank you for your helpful workflow and previous tips.
Quick question: On your RTX 4070 Super (12GB), how long does a standard 10-second video generation usually take with your V1.0 workflow using LTX + FP8?
Hi, I have 4070 TI not super for a 10 sec. video @1024 ~10min. using the "10Erosv1-fp8mixedlearned.safetensors" that is in the workflow.
How do I change the duration? How do I change it from 10 seconds to 20 seconds?
sadly, it did not run on my RTX4070 setup that meets the given requirements and config kept as imported (not changed settings, downloaded all models as given in the links) OutOfMemoryError on text encoding.
If I turn everything down to minimum still OOM error.
looking at the models that should be loaded, I cant see how this would fit a 12GB card by a long shot. "RTX4070" not at all. Or there are trick to this that are not stated. Wasted 2 hours on this damn.
Sorry that it did not work on your setup, all the videos that I have upload are made with this WF and my 4070Ti, Try this tips: First tip increase the paging file of windows to 122880MB (make a google search for how to do this, is easy), set this number to initial and maximum size, is better to use an SSD and even better a different SSD from that windows is installed. If that does not fix it, there is a red "Memory Saver" node turn both 1st and 2nd memory saver to ON. Finally edit your "run_nvidia_gpu.bat" with e text editor and add this "--lowvram --windows-standalone-build --use-sage-attention --windows-standalone-build --listen 0.0.0.0 --port 8188" after the "main.py". Hope that helps happy video generating :-), btw how much RAM do you have???
Any way I can get this to work with a 3080ti? LTX2MemoryEfficientSageAttentionPatch is throwing an error.
Hi, try to set "triton_kernels" to false in the "Patch Sage Attention KJ & Triton" subgraph. If that is not working, open the "Patch Sage Attention KJ & Triton" subgraph and bypass the "LTX2 Mem Eff Sage Attention Patch" node. Finally you can bypass the "Patch Sage Attention KJ & Triton" subgraph.
@panosgr this is a very good workflow man, thank you for the effort you put into this.