lichao-sun / Mora

Mora: More like Sora for Generalist Video Generation
1.47k stars 93 forks source link

Getting Very Bad Quality outputs #14

Open Deadsec69 opened 2 months ago

Deadsec69 commented 2 months ago

I was able to run the code exactly Somehow getting bad quality outputs and are not matching with the repo results These are some outputs I got for Text-conditional image-to-video generation Using the exactly same image mentioned in the results table

Prompt given: An image of a realistic cloud that spells “SORA”. https://github.com/lichao-sun/Mora/assets/27695803/c891546e-cdec-4801-806e-dac59f0e23d5

input2

Prompt given: Monster Illustration in the flat design style of a diverse family of monsters. The group includes a furry brown monster, a sleek black monster with antennas, a spotted green monster, and a tiny polka-dotted monster, all interacting in a playful environment. input1

https://github.com/lichao-sun/Mora/assets/27695803/fb9a0190-dbfe-4158-a895-2fc98a04efec

I am using stable-diffusion-v1-4 as mentioned in the code as well as OllamaLLM

I guess some model or some weights have been misssing😅

jagan607 commented 2 months ago

Hey @Deadsec69 , What's the GPU size it takes to run the repo?

Deadsec69 commented 2 months ago

I am using 80GB A100 GPU for videos and 80 GB another GPU to host olama @jagan607

atdvfegerf commented 2 months ago

@Deadsec69 can you please tell me how you managed to execute the code?