I’m looking for a better workflow to uprez video footage. I am often generating footage at a lower resolution.
Current workflow is not so sophisticated. Generate at lower res, put it through topaz to upscale it. I do not like the results I get running video through topaz. I often dont like the decisions it makes on chunkier parts of the image.
I’ve tried running stills through Magnific, then creating the video but haven’t been able to generate high resolution either. Agree on Topaz.
Yeah Topaz is not the ticket for me. Plus it’s quite expensive. Doing image generation > Magnific > imgtovideo generation within a model is pretty good.
What I’ve found remarkably recently is how high quality the videos out of Google Veo 2 are. They have minimal AI blemishes or warping.
Is Veo out for regular users now?
I don’t believe so. A google DeepMind researcher came to some of our meetups. They gave a few of us access. I could ask for you.
I can see him in over my head here already, maybe I’ll take pity on me. Trained photographer did college in the middle of switch over from film to digital for most folks, didn’t shoot commercially really outside of grad school, but kept up a little bit with a more creative and do things but I can color balance something all right. Been getting it at all of this AI generative thing, I’m doing OK with the steel images I haven’t tried much with video outside of topaz , and resolve. I have some old footage of my bands shows from 2000 or so and some of the videos really good but it’s all dark so I gotta come up a lot to really enjoy. So far, my brainstorm has left me with video, upscale with character model as a solution. I have well exposed quality photographs of everyone in the after party footage, and only two of us are really necessary for the upscale to be affective. Does anyone care to put in the direction of where to start with this kind of plan build a video lora maybe I can get by with a character, then it gets ran through something to break each frame into the still and upscaled generatively. Put it back together. I’ve been using comfyui locoly, I’m gonna do this video thing online I think. , right now I’m on osx I know it’s gonna be horrible waiting. I’m thinking I figure out how I’m gonna use I found the web interface find out what kind of model I need to make and I think I could take it from there. I just need suggestions on expectations in want tools to look at here?. I don’t really need free options. I don’t think you’re gonna get it done. I need them whatever they’re using make it look like Justin Bieber still alive.