Supermachine is an AI art generator bringing accessibility to AI art. No mentions of Google Collab here, no mentions of code, just a super comfy UX for using the best AI art models in a simple way.
Not sure if this is the right place to ask but I'll ask anyway. How to decide which resolution to choose for a generation? I've been always using the largest number in a fear that I will generate an awesome outcome but it will be unusable, and I won't be able to get the same result again. If you could elaborate on this, please. Much appreciated. Thanks.
Can I upload multiple pictures of the same person from different perspectives to provide better training? Can it be used for imaginary creatures, animals or only humans?
Thanks - this helped. I finally figured it out. An additional question though: say I would like to transmit an additional information (e.g. a name of a character) so I can later match it back in a database: how would I set this up?
Supermachine seems to work really well when the focus of the image is a person, how does it work with inanimate objects such as furniture, buildings and vehicles? Is there a specific model to use for inanimate objects? Can you create a video to show me how to create images that focus on inanimate objects?
This is so dope! But what if I wanna be funny and put my face on every person in a photo of 5 people? Also, does this work on anime or cartoonish figures?
consistent character is nice but is it possible to use a custom character like for example I want my daughters photo to be use as a reference character and do something similar to this video.
Hello Staurt,it would be useful if the "Image to Prompt" would include a suggested model. It will be super useful. Hope you consider adding this in future updates. Thank you.
Does this offer designs for logos, icons, typography, etc. that are more suitable for tee-shirts, mugs, door mats, banners, shower curtains, leggings, and other print-on-demand products? I don't just want artwork imagery. I want designs. Thanks!
Great video! Thanks a lot. I didn't notice the strength slider. 🙂 What I'm missing is a straightforward method to create icons for applications or mobile apps, particularly for generating business-related icons such as "settings," "add profile," and "delete." It would be fantastic if there was a tool designed for this purpose. Additionally, it would be excellent if users could download these images in SVG format.
@@udo9999 We did a very long video here - ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-LPcyuEUNUZc.html with all of the parts. We are releasing all of the sections in 4K over the next few weeks. All of the sections are in the larger video.
@@chillwalker of all the tools i've bought this is probably one of the worst ones lol. just butt ugly pictures that any free ai tool could do. There's a reason you can't make a refund for this trash
I recently purchased an impressive product and am interested in its integration capabilities. Specifically, I would like to know if there is a way to integrate our own models, such as those from Citivai AI. Is this feature currently available, or is it still under development? Additionally, I've noticed occasional inconsistencies in generated images, particularly concerning the eyes and hands, which can appear differently or somewhat imperfect. This issue also arises in photorealistic group images where only one image turns out well, while the rest, especially the background, appear distorted or not in sync. To address these issues, I suggest adding a feature for eye correction and an enhancement tool for group images. This would substantially improve both the quality and the consistency of the outputs. Furthermore, incorporating an option to upscale images during processing could help to maintain image quality, eliminating the need to enhance the images repeatedly over time. This extension could be a valuable addition to the enhancement features.
Unfortunately, so far all the updates have been for Supermachine's own models, since I use SDXL almost exclusively, I haven't benefited from any updates so far, which I unfortunately find to be a shame and one-sided in terms of development.
Was really hoping that the implementation of ControlNet's Open Pose on Supermachine would let users have the freedom to create bespoke poses or generate from image references? :(
This is incredible! What I'd really love to do though is to create my own consistent character. For example, I've an illustrated cartoon mascot squirrel I would like to have consistently output into different poses or scenes (such as pointing or waving). Is this something we might be able to do eventually with Supermachine?
I want to generate my own custom consistent characters though. I don’t want to use characters that anyone else is using. I want to train the AI with my own characters.
4:26 This is cool, but is there a way to apply a specific art style to the post-apocalyptic image of London that you created? For example, can we turn this image into an anime-style background?
Is there a way to take a consistent character and depict them in different ages? For example, I might want to create scenes where I go back in time and show that character as a kid. In other cases, I might want to go forward in time and show that character in their old age. Once I decide that I want to depict a character in a specific age, I also need that character to be consistently depicted in that age in every scene (until I decide to show them in a different age). This is important for storytelling.
Consistent Characters 18:52 Can we train Supermachine to create our own consistent characters? I don’t want to use consistent characters that everyone else is using. I want to use my own custom consistent characters, based on unique training images of my character that I choose.
