My god SFX is going crazy everyday. when compared to version 4.1 back then and now in 2023. There is lot of changes that boris is doing. great job and all the efforts are coming together guys. cheers and looking for more stuff soon.
Are you using the Silhouette plugin inside Premiere Pro? If so, it works exactly the same except you have to apply the Silhouette plugin to a clip first and then open up the Silhouette UI. If that doesn't answer your question, please let us know with some more details about what you want to do.
I run my stable diffusion locally, not only is it better since I have more custom models, but it also makes more sense since no credit limit. Will this work with locally-run stable diffusion on Automatic1111?
Do we know anything about the resolution of the images generated by StabilityAi? More credits for higher res? I’ve used Dall-e for similar workflow, but it is resolution locked and HD sizes. Nice results tho.
Yes, there are four modes for Max Size. The most common one is Auto which automatically sets the Max Size based on the session size. For sessions larger than 1024 pixels, the image is scaled down to the to 1024 limit and the result is scaled back up to the session size. Size and Steps will change the speed to produce and credits used to generate the images. All generated images are stored with the project.
I followed this tutorial trying to remove a simple object in the back and my results were horrendous. What I got instead of the lamp being removed I got a big blurry image of something. What can I be doing wrong?
Check that you are feeding a correct alpha channel into Stability. That's the usual issue. In your Input node, you may have to change the Fill dropdown to "None". Any other problems, please join us on the forums as it's easier to share screenshots there : forum.borisfx.com/
Do you have your API key copied into Preferences and is your drive writeable? This sounds like a permissions issue, but please contact Support so they can help troubleshoot further : support.borisfx.com/hc/en-us/requests/new
It's using Stability.AI and creating full res images. Not samples. You decide how many images it creates. Depending on your settings each image costs about $0.01-$0.02. Just to underline this, Boris FX is not charging you for creating images and makes no money doing so. This just makes an easier workflow for those who want to use Stability.ai's system to generate assets or test ideas. EDIT : New pricing brings the actual cost to about $0.002 (0.2c) for an SDXL1.0 images at 20 steps.
My god SFX is going crazy everyday. when compared to version 4.1 back then and now in 2023. There is lot of changes that boris is doing. great job and all the efforts are coming together guys. cheers and looking for more stuff soon.
It's wonderful to see a "pre-nodal" Silhouette user. Things were very different in those days. The changes have been huge. Thanks!
Excellent originality This is one of my favourite Boris FX tutorials out of the many that I have seen.
Thanks so much for the great comment.
Awesome! Cant wait to dig into this! Ready for part 2 of this one!
Silhouette is fast becoming the most fun software for compositing for me! 😍
That's really fantastic to hear. Thanks for the comment.
wow. that thing can do anything
Well I used to remove object using the mocha pro plugin only. Didn't actually know silouette could do the same.
Yes, it's very flexible. (You even have Mocha Pro built directly in to Silhouette when you need a regular Mocha remove!)
Very well explained - thank you!
You're very welcome! Thanks for watching and taking the time to leave a comment.
Awesome tutorial!
worth a try.
Thanks!
Interesting channel! How to save and transfer changes in premier pro?
Are you using the Silhouette plugin inside Premiere Pro? If so, it works exactly the same except you have to apply the Silhouette plugin to a clip first and then open up the Silhouette UI. If that doesn't answer your question, please let us know with some more details about what you want to do.
I run my stable diffusion locally, not only is it better since I have more custom models, but it also makes more sense since no credit limit. Will this work with locally-run stable diffusion on Automatic1111?
No, we don't support local Stable Diffusion installs. Stability offers an API, which makes this a more consistent and supportable experience.
It's on the list!
@@PaulMillerFx go go go!
Do we know anything about the resolution of the images generated by StabilityAi? More credits for higher res? I’ve used Dall-e for similar workflow, but it is resolution locked and HD sizes. Nice results tho.
Yes, there are four modes for Max Size. The most common one is Auto which automatically sets the Max Size based on the session size. For sessions larger than 1024 pixels, the image is scaled down to the to 1024 limit and the result is scaled back up to the session size. Size and Steps will change the speed to produce and credits used to generate the images. All generated images are stored with the project.
I followed this tutorial trying to remove a simple object in the back and my results were horrendous. What I got instead of the lamp being removed I got a big blurry image of something. What can I be doing wrong?
Check that you are feeding a correct alpha channel into Stability. That's the usual issue. In your Input node, you may have to change the Fill dropdown to "None".
Any other problems, please join us on the forums as it's easier to share screenshots there : forum.borisfx.com/
I'm having trouble when i ask to Generate few seconds after pop up this mensage "Could not create a file", anyone seen something like before?
Do you have your API key copied into Preferences and is your drive writeable? This sounds like a permissions issue, but please contact Support so they can help troubleshoot further : support.borisfx.com/hc/en-us/requests/new
Mi mocha no quiere entrar video cuando lo entro no sube la imagen del vídeo por favor ayúdeme por favor 🙏🙏
Please give Support the message. They are the best people to help troubleshoot what the problem could be : support.borisfx.com/hc/en-us/requests/new
consuming credits even for generating for samples is completely unethical !! it should be only consumed after final confirmation to use.
It's using Stability.AI and creating full res images. Not samples. You decide how many images it creates. Depending on your settings each image costs about $0.01-$0.02.
Just to underline this, Boris FX is not charging you for creating images and makes no money doing so. This just makes an easier workflow for those who want to use Stability.ai's system to generate assets or test ideas.
EDIT : New pricing brings the actual cost to about $0.002 (0.2c) for an SDXL1.0 images at 20 steps.
Is there a planned way to be able to run stable diffusion locally instead via the cloud?
We are looking at different methods to extend the Stability node. No timelines available on this though.