ltx-2.3-spatial-upscaler distorts the results of the first generation step
Tried on workflows like FL2V and ic-lora (controlNet). After the first generation step, the last frame corresponds to the last frame, the movements correspond to the original video. After the second step, the final frame roughly corresponds to the prompt rather than the final frame. The coordinated dance movements turn into flailing arms and legs.
I strip the spatial and latent upscaling out completely. The upscaling process induces artifacts and body horror, and almost always changes the face of the subject. Once I stripped the upscaling out, these things never happen. Downscaling the image and then upscaling the video later takes the same amount of time as just generating the video at full resolution with none of the side effects. Also, to fix the slow-motion thing that LTX does, change your FPS to 25 instead of 24. I don't know why so many workflows set the FPS to 24 when the model was trained at 25 and 50. It supports 24 but you're going to get either slow or fast motion.
A lot of these workflows were made for LTX 2.0, so it stands to reason that they were set for 24 FPS. 2.3 handles 24, 25, 48 and 50 FPS.
Just because 2.3 can “handle” 24 fps doesn’t mean it should be used. Go up 1 frame per second and the slow motion issues stop. Simple.
Then send your complaints to the creators of the workflow templates. Not everyone uses the same workflow template, text encoders (stock vs Heretic), or models (stock vs Kijai) and that does have a major impact on the end result.
It doesn’t matter what models you use. LTX 2.3 was trained natively on 25 and 50, not 24 and 48. It can “handle them” but why go through that if you can just go up 1 fps to stop the slow mo thing. Just stay in your lane bro. No one asked you for your opinion.
Wow, you're a testy little child aren't you? Maybe you should take your aggression elsewhere, because it's not doing you any justice on these forums.
Whatever you say sonny boy. You’re just mad you’re wrong