Topic: Can I have some help finding

Posted under Off Topic

I don't know if it helps but I would like some help making Ai picture to video in a creative way I just hope I am not being a pest too anyone on this idea I just like to make pictures talk or say something that fits the theme as they move

gushcubbat said:
I don't know if it helps but I would like some help making Ai picture to video in a creative way I just hope I am not being a pest too anyone on this idea I just like to make pictures talk or say something that fits the theme as they move

Most of the creative capability that people have right now for making an AI work more personal either requires post-production skills befitting that of someone who is already a capable artist or just honing in on your prompts better unfortunately. Very little is actually that fine-grained right now and most of the time you are working on image generation things will start to get annoyingly sticky a couple of iterations in.

gushcubbat said:
I don't know if it helps but I would like some help making Ai picture to video in a creative way I just hope I am not being a pest too anyone on this idea I just like to make pictures talk or say something that fits the theme as they move

Always great to see more people willing to try out genning! If you're interested in doing image to video gens, you can either use an online service or do it on your own computer. Online services might need memberships or payment, while on your own computer you'll need a decent GPU.

Like LemurPrincess said, for your own computer using ComfyUI is probably your best option. The two most popular modern local image video gen models are LTX 2.5 and Minimax H3. You can find tutorials for them at https://comfy.org/minimax-h3/ and https://docs.comfy.org/tutorials/video/ltx/ltx-2-5 .

Online video gen services are higher quality, but they're all censored. The most uncensored is Grok, but they've been censoring things more and more and video gens are no longer free: https://grok.com/imagine . Another popular site is https://higgsfield.ai/ which gives access to a lot of different models, but requires a membership and payment. Finally, Google sometimes lets you do free video gens through Google Flow, but again it's censored and Google's TOS says they can ban your Google account(s) if you try to evade censorship: https://flow.google.com/

Hope that helps! Whatever you decide to do, best of luck with it!

gridanon said:
Always great to see more people willing to try out genning! If you're interested in doing image to video gens, you can either use an online service or do it on your own computer. Online services might need memberships or payment, while on your own computer you'll need a decent GPU.

Like LemurPrincess said, for your own computer using ComfyUI is probably your best option. The two most popular modern local image video gen models are LTX 2.5 and Minimax H3. You can find tutorials for them at https://comfy.org/minimax-h3/ and https://docs.comfy.org/tutorials/video/ltx/ltx-2-5 .

Online video gen services are higher quality, but they're all censored. The most uncensored is Grok, but they've been censoring things more and more and video gens are no longer free: https://grok.com/imagine . Another popular site is https://higgsfield.ai/ which gives access to a lot of different models, but requires a membership and payment. Finally, Google sometimes lets you do free video gens through Google Flow, but again it's censored and Google's TOS says they can ban your Google account(s) if you try to evade censorship: https://flow.google.com/

Hope that helps! Whatever you decide to do, best of luck with it!

can i dm you if i have questions or just try the tutorials?

i tired them both they didn't hit me right so i guess i stick with one site that hits it right enough but i do hope people share more links and options

Gushcubbat reached out to me in DMs for more info on how I do videos, but I wanted to post it publicly in case anyone else finds it helpful.

The highest quality videos are all going to be from corporate models. Unfortunately those are censored and typically paid-only, although sometimes on some sites you can get free trials.

Examples of corporate model videos include https://blep.art/posts/142364 , https://blep.art/posts/142883 and https://blep.art/posts/150300 . Those were all make by the Sora 2 model which is no longer available, but models like Seedance 2.5, Kling 3.0, Google Veo, and Grok video can reach the same or greater quality. Grok used to be the most uncensored but has gotten more censored as time goes on.

One of the most popular sites for AI video gens with these models is Higgsfield: https://higgsfield.ai/ . There's a ton of tutorials on youtube for any site and any model, here's one: https://www.youtube.com/watch?v=Mt91SiQotOU

Alternatively, the genner Hellhound Rhythms has done a ton of video and has even gotten uncensored results. He uses https://imaginex.video/ mainly Grok Imagine and their Ultra Open model for NSFW.

Unfortunately those are all paid options. If you want to do video gens for free, you'll probably need to use a local model on your PC. Local models usually can't reach the same quality as corporate models, but can still be impressive. Examples of free local model gens include https://blep.art/posts/140428 , https://blep.art/posts/196206 , and https://blep.art/posts/172146 .

Probably the best local UI for video gens is ComfyUI. You can get it at https://comfy.org/ .

The best modern video models for local AI gens are Wan2.2, LTX 2.5, and Minimax H3. WAN 2.2 is the oldest but might be the best uncensored model before any LORAs, Minimax H3 is the newest and what I'd recommend now but you'll probably need LORAs to get accurate explicit gens, and LTX 2.5 is somewhere in the middle.

You can find the ComfyUI official workflows and tutorials for each of those at:

https://docs.comfy.org/tutorials/video/wan/wan2_2
https://docs.comfy.org/tutorials/video/ltx/ltx-2-5
https://docs.comfy.org/tutorials/video/minimax/minimax-h3

In addition, there are a ton of video tutorials available on Youtube. Here's one: https://www.youtube.com/watch?v=qUFWJEJT77o

My personal recommendations are to start small with simple prompts at low resolutions and low quality settings because that will gen the fastest, and then you can keep making the prompt more complex to get the video closer to what you want and start increasing the quality settings once it's close to what you want.

Feel free to ask if anyone has any questions. Hope everyone gives video genning a shot if you can. Best of luck with all your gens!