Update README.md
Browse files
README.md
CHANGED
|
@@ -15,22 +15,22 @@ license_link: https://github.com/Lightricks/LTX-2/blob/main/LICENSE-2_x
|
|
| 15 |
library_name: comfyui
|
| 16 |
pipeline_tag: video-to-video
|
| 17 |
widget:
|
| 18 |
-
- text: "
|
| 19 |
output:
|
| 20 |
url: Examples/ltx_001.mp4
|
| 21 |
-
- text: "
|
| 22 |
output:
|
| 23 |
url: Examples/ltx_002.mp4
|
| 24 |
-
- text: "
|
| 25 |
output:
|
| 26 |
url: Examples/ltx_003.mp4
|
| 27 |
-
- text: "
|
| 28 |
output:
|
| 29 |
url: Examples/ltx_004.mp4
|
| 30 |
-
- text: "
|
| 31 |
output:
|
| 32 |
url: Examples/ltx_005.mp4
|
| 33 |
-
- text: "
|
| 34 |
output:
|
| 35 |
url: Examples/ltx_006.mp4
|
| 36 |
---
|
|
@@ -39,11 +39,7 @@ widget:
|
|
| 39 |
|
| 40 |
### Edit one frame. Ripple the change through the whole video.
|
| 41 |
|
| 42 |
-
LTX Ripple is a **First Frame All Frames (FFAF) IC-LoRA** for [LTX-2.5](https://huggingface.co/Lightricks/LTX-2.5). It
|
| 43 |
-
|
| 44 |
-
<p align="center">
|
| 45 |
-
<a href="https://huggingface.co/WepeNerd"><strong>Created by WepeNerd</strong></a>
|
| 46 |
-
</p>
|
| 47 |
|
| 48 |
<Gallery />
|
| 49 |
|
|
@@ -51,27 +47,26 @@ LTX Ripple is a **First Frame All Frames (FFAF) IC-LoRA** for [LTX-2.5](https://
|
|
| 51 |
|
| 52 |
1. Start with a video.
|
| 53 |
2. Edit or replace its first frame.
|
| 54 |
-
3.
|
| 55 |
-
4. LTX Ripple carries the first-frame edit through the video.
|
| 56 |
|
| 57 |
-
You
|
| 58 |
|
| 59 |
Useful applications include:
|
| 60 |
|
| 61 |
-
- Special
|
| 62 |
-
-
|
| 63 |
-
-
|
| 64 |
-
-
|
| 65 |
-
-
|
| 66 |
-
-
|
| 67 |
-
-
|
| 68 |
-
-
|
| 69 |
-
-
|
|
|
|
|
|
|
| 70 |
|
| 71 |
## Examples
|
| 72 |
|
| 73 |
-
The gallery above shows the six included examples. Each caption describes the first-frame edit and the effect that Ripple propagates through the video.
|
| 74 |
-
|
| 75 |
## Files
|
| 76 |
|
| 77 |
| File | Purpose |
|
|
@@ -98,28 +93,8 @@ The workflow is built for LTX-2.5 and references the following model components:
|
|
| 98 |
- `gemma4-12b-with-proj-ltx-2.5-comfy-int8-convrot.safetensors`
|
| 99 |
- `gemma4_e2b_it_int8_convrot.safetensors` — optional, only when prompt enhancement is enabled
|
| 100 |
|
| 101 |
-
|
| 102 |
-
|
| 103 |
-
- [ComfyUI-LTXVideo](https://github.com/Lightricks/ComfyUI-LTXVideo)
|
| 104 |
-
- [ComfyUI-KJNodes](https://github.com/kijai/ComfyUI-KJNodes)
|
| 105 |
-
- [RES4LYF](https://github.com/ClownsharkBatwing/RES4LYF)
|
| 106 |
-
- [ComfyUI-VideoHelperSuite](https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite)
|
| 107 |
-
- [cg-use-everywhere](https://github.com/chrisgoringe/cg-use-everywhere)
|
| 108 |
-
|
| 109 |
-
## Recommended settings
|
| 110 |
-
|
| 111 |
-
The included workflow uses the following starting point:
|
| 112 |
-
|
| 113 |
-
| Setting | Starting value |
|
| 114 |
-
|---|---:|
|
| 115 |
-
| LoRA model strength | `1.35` |
|
| 116 |
-
| Frame rate | `25 fps` |
|
| 117 |
-
| Video length | `97 frames` |
|
| 118 |
-
| Guide frame index | `0` |
|
| 119 |
-
| Guide strength | `1.0` |
|
| 120 |
-
| Resolution | `1152 × 768` in the video-input branch |
|
| 121 |
-
|
| 122 |
-
These are starting points, not hard requirements. Increase the LoRA strength when the edit is not being carried through strongly enough. Decrease it when the output changes content that should remain unchanged.
|
| 123 |
|
| 124 |
## Prompting
|
| 125 |
|
|
|
|
| 15 |
library_name: comfyui
|
| 16 |
pipeline_tag: video-to-video
|
| 17 |
widget:
|
| 18 |
+
- text: "Example of replacing an object, adding special effects, and changing the lighting all at the same time. The scene reacts to both new light sources."
|
| 19 |
output:
|
| 20 |
url: Examples/ltx_001.mp4
|
| 21 |
+
- text: "Example of changing the lighting and background of the first frame, and the rest of the video is figured out without prompting, even when the camera swivels around."
|
| 22 |
output:
|
| 23 |
url: Examples/ltx_002.mp4
|
| 24 |
+
- text: "Editing the first frame to be on fire, the effect stays consistent throughout multiple camera cuts."
|
| 25 |
output:
|
| 26 |
url: Examples/ltx_003.mp4
|
| 27 |
+
- text: "Example of Ripple understanding that the liquid has been added in the first frame, and the liquid inside the glass flows in accordance with the original video's movement."
|
| 28 |
output:
|
| 29 |
url: Examples/ltx_004.mp4
|
| 30 |
+
- text: "Example of archviz/3d clay model texturing with only the untextured video and a textured first frame input. Texturing effect ripples through scene cuts."
|
| 31 |
output:
|
| 32 |
url: Examples/ltx_005.mp4
|
| 33 |
+
- text: "Here an interesting effect can be seen: because the edited first frame was sharper (notice the dragon details), the entire video has been sharpened."
|
| 34 |
output:
|
| 35 |
url: Examples/ltx_006.mp4
|
| 36 |
---
|
|
|
|
| 39 |
|
| 40 |
### Edit one frame. Ripple the change through the whole video.
|
| 41 |
|
| 42 |
+
LTX Ripple is a **First Frame All Frames (FFAF) IC-LoRA** for [LTX-2.5](https://huggingface.co/Lightricks/LTX-2.5). It allows you to make LTX as powerful as your favorite image editor, with smart first frame propagation that understands the edit contextually.
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
|
| 44 |
<Gallery />
|
| 45 |
|
|
|
|
| 47 |
|
| 48 |
1. Start with a video.
|
| 49 |
2. Edit or replace its first frame.
|
| 50 |
+
3. LTX Ripple will figure out the rest of the video.
|
|
|
|
| 51 |
|
| 52 |
+
You may also give a brief description of what you changed to help it along, but the LoRA can work without even prompting anything.
|
| 53 |
|
| 54 |
Useful applications include:
|
| 55 |
|
| 56 |
+
- Adding Special Effects
|
| 57 |
+
- 3D model re/texturing
|
| 58 |
+
- Style Change
|
| 59 |
+
- Light source / time of day change
|
| 60 |
+
- Character Attribute Change
|
| 61 |
+
- Background Change
|
| 62 |
+
- Weather Change
|
| 63 |
+
- Archviz Retexturing
|
| 64 |
+
- Title/Logo Replacement
|
| 65 |
+
- Object Addition / Removal / Replacement
|
| 66 |
+
- Sharpening / Blurring
|
| 67 |
|
| 68 |
## Examples
|
| 69 |
|
|
|
|
|
|
|
| 70 |
## Files
|
| 71 |
|
| 72 |
| File | Purpose |
|
|
|
|
| 93 |
- `gemma4-12b-with-proj-ltx-2.5-comfy-int8-convrot.safetensors`
|
| 94 |
- `gemma4_e2b_it_int8_convrot.safetensors` — optional, only when prompt enhancement is enabled
|
| 95 |
|
| 96 |
+
## Recommended strength setting: 1.3
|
| 97 |
+
Increase the LoRA strength when the edit is not being carried through strongly enough. Decrease it when the output changes content that should remain unchanged.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
|
| 99 |
## Prompting
|
| 100 |
|