this is what I was looking for but...
I am working on a personal short film, and as for me, the worst thing with LTX is to have only a first reference image where we need to put everything with Flux or Qwen... I am trying to test your pipeline, and thank you for providing sample images and final videos, but too bad they were not the prompts to achieve it .... Could you publish them? It would be helpful. Thank you.
I am working on a personal short film, and as for me, the worst thing with LTX is to have only a first reference image where we need to put everything with Flux or Qwen... I am trying to test your pipeline, and thank you for providing sample images and final videos, but too bad they were not the prompts to achieve it .... Could you publish them? It would be helpful. Thank you.
I didn't post the prompt because I wrote in chinese, I'll post the translated prompt here.
sample1:At the entrance of a flower shop, the shot first reveals a three-tier wooden flower stand and a light apricot-colored wall. An orange-and-white shorthair cat peeks out from beside a flower bucket, then lowers its head to sniff the edge of a bouquet wrapped in kraft paper. The camera gently pushes in at the height of the flower stand, letting the cat pause on one side of the threshold. The morning light of the flower shop blends with the greenery, creating a soft 3D texture.
sample2:On a pine-forest mountain trail in the morning, the shot first reveals a dirt-road bend and the rising shadow of a small aerial drone. A mountain biker in a fluorescent yellow cycling outfit rides a black-and-orange full-suspension mountain bike, leaning into the turn and rushing across the gravel path. The camera follows quickly from a low angle, capturing the wheels splashing mud and the rider leaning sideways. The live-action dappled sunlight creates an exciting and refreshing atmosphere.
Thank you very much. I have made tests during the night with your LORA. It's very promising.
I tried to inject two reference images for my main character, one for the head and one for the whole body, but the result does not follow the original design much. The character looks short and fat— the limb problem that you mention in the readme file. But animation is very realistic and fluid without artifacts or hallucinations.
Ideally, your system combined with Prompt Relay, and we could have a tool ready to produce cinematographic sequences. Do you think it's possible to mix both workflows? I will try to see by my side, but my knowledge level in ComfyUI coding is too bad.
Thank you very much. I have made tests during the night with your LORA. It's very promising.
I tried to inject two reference images for my main character, one for the head and one for the whole body, but the result does not follow the original design much. The character looks short and fat— the limb problem that you mention in the readme file. But animation is very realistic and fluid without artifacts or hallucinations.
Ideally, your system combined with Prompt Relay, and we could have a tool ready to produce cinematographic sequences. Do you think it's possible to mix both workflows? I will try to see by my side, but my knowledge level in ComfyUI coding is too bad.
It seems that some people in the community have already combined the two workflows successfully. You may be able to find examples by searching around.
Also, thank you very much for your feedback. It’s very helpful to hear your testing results.