The current inpainting canvas tool doesn't work, at all. - The original image is lost, the prompt completely reinvents the image! - We cannot just upload an image there, we have to first use img2img, then open canvas. - The Canvas window is way too small (we don't see the full image). - It's also very slow to generate (4 to 5 minutes) - Tools like remove background should be made avaible the canvas.
A: Hey Sebastien, at this time we are not looking to add Framepack as it is multi minute generations for a short video ( a few seconds ) compared to even longer for a longer video.
In terms of updates, please login to your account and you would see multiple updates over the past months. Here is a selection of them;
- Added the Lumina 2 model which has a really interesting capability for sketch...
Framepack offers two minutes long videos. We need some kind of announcement popup or release notes, when we login we don't even see the newer models it shows older models by default. Now that you told me indeed found these... The UI really deserves improvement.
The Gemini instruct is not available on my UI. All these tools also require a better UI implementation, merging them together.
This is no joke. Framepack takes hours to produce a video. I can't imagine the strain it would put on the super machine network.
Q: Will this eventually have human motion?
Hi, I do not know much about this technology, so apologies if the question shows a bit of ignorance.
I see fake models on Instagram that are AI generated. They often have the model moving, which makes it a video. Will this eventually be able to do the same?
Adding human motion, or making a short video, opens up use on TikTok, which is my main reason for asking.
My guess is that a real human is making the movement and then they are somehow overlaying the AI image, but I don't know.
A: Hey there, human motion is a lot more resource intensive to run and tends to take multiple minutes to generate for a single 5 second clip. We see the technology needs a lot of improvement before it could be put into a product like Supermachine. A lot of the video generations at the moment are done on expensive hardware by individuals which is why you don't see many specific video tools. We'd...
Framepack is fun to play with but the quality is below the threshold of what I'd deem of worthy for a video model that is super slow to create a video so we don't plan to implement it at this time. It wouldn't be a good experience if users are waiting 5-10 mins for a short video.
Q: T-shirt Designs
How do I incorporate this to get designes create for T-shirts?
Q: inpainting canvas not working
The current inpainting canvas tool doesn't work, at all.
- The original image is lost, the prompt completely reinvents the image!
- We cannot just upload an image there, we have to first use img2img, then open canvas.
- The Canvas window is way too small (we don't see the full image).
- It's also very slow to generate (4 to 5 minutes)
- Tools like remove background should be made avaible the canvas.
Any plan to fix these and build a v2.0 UI?
Share SUPERMACHINE
Q: Can we download a picture or video of ourselves or a pet for example?
Can we upload a picture or video of ourselves or a pet, for example, and use it in the product? Is there an additional cost for this?
Share SUPERMACHINE
Q: Questions
Can we have this new free unsensored tool for video generation:
https://youtu.be/3eoUoPtLMPI?feature=shared
Can we have multiple items uploaded in images? Most other tools offer that capability now (product presentation, custom background etc)
I didn't see any update to the tool for the past 5 months, where do you publish your release notes and roadmap?
Rexxar
Apr 28, 2025A: Hey Sebastien, at this time we are not looking to add Framepack as it is multi minute generations for a short video ( a few seconds ) compared to even longer for a longer video.
In terms of updates, please login to your account and you would see multiple updates over the past months. Here is a selection of them;
- Added the Lumina 2 model which has a really interesting capability for sketch...
Share SUPERMACHINE
Verified purchaser
Framepack offers two minutes long videos.
We need some kind of announcement popup or release notes, when we login we don't even see the newer models it shows older models by default. Now that you told me indeed found these... The UI really deserves improvement.
The Gemini instruct is not available on my UI. All these tools also require a better UI implementation, merging them together.
Verified purchaser
This is no joke. Framepack takes hours to produce a video. I can't imagine the strain it would put on the super machine network.
Q: Will this eventually have human motion?
Hi, I do not know much about this technology, so apologies if the question shows a bit of ignorance.
I see fake models on Instagram that are AI generated. They often have the model moving, which makes it a video. Will this eventually be able to do the same?
Adding human motion, or making a short video, opens up use on TikTok, which is my main reason for asking.
My guess is that a real human is making the movement and then they are somehow overlaying the AI image, but I don't know.
Rexxar
Mar 27, 2025A: Hey there, human motion is a lot more resource intensive to run and tends to take multiple minutes to generate for a single 5 second clip. We see the technology needs a lot of improvement before it could be put into a product like Supermachine. A lot of the video generations at the moment are done on expensive hardware by individuals which is why you don't see many specific video tools. We'd...
Share SUPERMACHINE
Verified purchaser
The author of ControlNet recently created something really cool, and I think we should pay attention to it.
Verified purchaser
You are FAR more advanced in this area than I am. I have no idea what ControlNet is, but thanks for the heads up.
Verified purchaser
ControlNet is currently most commonly used for image-to-image generation. Supermachine also utilizes it.
Verified purchaser
By the way, this is what I mentioned at the beginning.
https://github.com/lllyasviel/FramePack
Framepack is fun to play with but the quality is below the threshold of what I'd deem of worthy for a video model that is super slow to create a video so we don't plan to implement it at this time. It wouldn't be a good experience if users are waiting 5-10 mins for a short video.
Q: T-shirt Designs
How do I incorporate this to get designes create for T-shirts?
Share SUPERMACHINE