This is great! The only thing is... and maybe I'm an idiot, but I can't find anything about how to get this AI plugin? When I go to the menu specified in the other video, there's no AI option, so I figure it's a plugin, but I can't find that anywhere out there?
I may put this on GitHub if I have the time but it would require some know how. It's actually just an easy to make a build that can be readily deployed, but the problem is that I can't release it because OpenAI API is not free.
@@Hdhead Thanks for your answer. I understand that. A tutorial on how to get open ai into resolve in general for this purpose would be interesting if publishing the project is not possible.
Hi there Igor! I also tried running some scripts with python from Workspace > Scripts > Comp > 5_get_project_information That's the python file in the Examples folder, but nothings happening. i also followed all of the instructions in the README file Are there any debugging tools for this? I don't know why it's not working
Great video. Question - is it possible to modify the ChatyGPT script to interface with DALL-E to use the generate AI to modify/patch part of the image in the footage identified by mask? For example, let's say I have a footage of a beautiful city street but I like to remove/patch up a big trash can in part of the footage. Using the Resolve Object Remover OFX effect is time consuming and gets less than optimal results, can I have DALL-E do it? Thanks.
The Resolve's object remover is engineered to work with moving images. DALL-E is not. DALL-E has an inpainting feature, but you'd still need a good amount of manual skills to integrate the clean patch into the plate.
@@Hdhead Cool, but you should ditch Dall-E and go for an open (and more ethical) model like Stable Diffusion instead. Something like like the Krita extension here (perhaps using fusion for inpainting) would be more powerful, ruclips.net/video/h-BIC9iuPGc/видео.html and does not depend on the whims of a centralized service.
@@Hdhead Thanks for the explanation. I hope either DALL-E or Runway can do the object remover via Generate AI on moving video soon, as that will be 100 times more useful and powerful than just generating pictures based on text description.
@@thebrotherhood-podcast Yes, I may do that. For anyone who is already doing Resolve scripting the OpenAI API is super simple. The only value I added at this time is the scope limiting. That's too experimental to release.
If you mean "inpainting" that's possible with the OpenAI API, but I didn't think that would of interest since there is a built in tool for that In Resolve. If you just want to automate placing a cutout of a DALL-E generated image over an existing image, the Resolve API doesn't have anything like that, but I could cobble up a way to do it.
It's great... But what did I benefit from your offer if I didn't use those tools.. Is this, for example, a video showcasing your skills or what? It would have been better to present the tool to the followers, or at least explain how it works.
Good question. I write tools for Resolve. I'm putting this out to see if people think it could be useful or if people have other ideas for AI powered tools.
@@Hdhead Part of the original idea was to not only to have coders to contribute but to have a community uploading data and and running models on servers to aid in training. I have about 2TB of just facial images that I don't even remember which drive it is on because I quit messing with it when 'OPEN' went 'PRIVATE'. Look at DRUPAL ... that was almost entirely community developed.
This is amazing I definitely see how this can be used to speed up workflows
Can't wait for the fusion one
so interesting Igor, thanks.
This is great!
The only thing is... and maybe I'm an idiot, but I can't find anything about how to get this AI plugin? When I go to the menu specified in the other video, there's no AI option, so I figure it's a plugin, but I can't find that anywhere out there?
It's not a plugin. It's something I wrote and it's not released to the public. It's a working proof of concept.
@@HdheadAh ok, thanks. It seems to be excellent! Keep up the good work!
Great for storyboarding!
Dope
That's cool 🫡..waiting for the part 3 !
Absolutely crazy stuff
Really amazing, is there a way to try this out? Would be really cool to see a tutorial about how to get this up and running!
I may put this on GitHub if I have the time but it would require some know how. It's actually just an easy to make a build that can be readily deployed, but the problem is that I can't release it because OpenAI API is not free.
@@Hdhead Thanks for your answer. I understand that. A tutorial on how to get open ai into resolve in general for this purpose would be interesting if publishing the project is not possible.
Hi there Igor! I also tried running some scripts with python from Workspace > Scripts > Comp > 5_get_project_information
That's the python file in the Examples folder, but nothings happening. i also followed all of the instructions in the README file
Are there any debugging tools for this? I don't know why it's not working
There is a lot of helpful discussion on the BMD official forum.
It is amazing ... Michael Jackson and Glenn Glove! 🙂
Do you have the scripts available to play around with?
I have Resolve scripts on my GitHub, but not these OpenAI based ones.
where did you learn programming for Resolve I want to get also into it but I don't know where to start
I learned it from the API README document that ships with Resolve.
Great video. Question - is it possible to modify the ChatyGPT script to interface with DALL-E to use the generate AI to modify/patch part of the image in the footage identified by mask? For example, let's say I have a footage of a beautiful city street but I like to remove/patch up a big trash can in part of the footage. Using the Resolve Object Remover OFX effect is time consuming and gets less than optimal results, can I have DALL-E do it?
Thanks.
The Resolve's object remover is engineered to work with moving images. DALL-E is not. DALL-E has an inpainting feature, but you'd still need a good amount of manual skills to integrate the clean patch into the plate.
@@Hdhead Cool, but you should ditch Dall-E and go for an open (and more ethical) model like Stable Diffusion instead. Something like like the Krita extension here (perhaps using fusion for inpainting) would be more powerful, ruclips.net/video/h-BIC9iuPGc/видео.html and does not depend on the whims of a centralized service.
@@Hdhead Thanks for the explanation. I hope either DALL-E or Runway can do the object remover via Generate AI on moving video soon, as that will be 100 times more useful and powerful than just generating pictures based on text description.
Fannntastic!!!! When will you release this for public?
I don't know. I'm trying to see what the interest for something like this would be. It would not be free--OpenAI charges for the API use.
@@Hdhead You could release the source code without the API keys and have people use their own
@@thebrotherhood-podcast Yes, I may do that. For anyone who is already doing Resolve scripting the OpenAI API is super simple. The only value I added at this time is the scope limiting. That's too experimental to release.
Wow my brainblow bro 🤩
have a way to make a mask and fill with a ai image?
If you mean "inpainting" that's possible with the OpenAI API, but I didn't think that would of interest since there is a built in tool for that In Resolve. If you just want to automate placing a cutout of a DALL-E generated image over an existing image, the Resolve API doesn't have anything like that, but I could cobble up a way to do it.
OOOOOOOOH NICE!!!
woooow
It's great... But what did I benefit from your offer if I didn't use those tools.. Is this, for example, a video showcasing your skills or what? It would have been better to present the tool to the followers, or at least explain how it works.
Good question. I write tools for Resolve. I'm putting this out to see if people think it could be useful or if people have other ideas for AI powered tools.
It's so cool !!!!!!! You can animate it and play take on me 🎹
Aha.
😂
Interesting but not interested ~ OpenAI should change its name to 'Now making money on the open work of others'.
I doubt they are making any money yet. The compute cost of the training and running the models is astronomical.
@@Hdhead Part of the original idea was to not only to have coders to contribute but to have a community uploading data and and running models on servers to aid in training. I have about 2TB of just facial images that I don't even remember which drive it is on because I quit messing with it when 'OPEN' went 'PRIVATE'. Look at DRUPAL ... that was almost entirely community developed.