Image to image
planned
Merlin
planned
F
Felipe Matos
This will be very valuable!
Suggested Features:
Image-to-Image (img2img):
Ability to use existing images as reference for new generations
Option to maintain structural elements while changing styles or contents
Control over how closely the output follows the input image
Inpainting:
Selective editing of specific areas in an image
Ability to remove unwanted elements or background
Option to add new elements while maintaining image consistency
Benefits:
Increased workflow flexibility for creative professionals
More complete editing suite within Merlin.ai
Reduced need for external editing tools
Enhanced creative control over generated images
Better iteration capabilities for design processes
Use Cases:
Designers refining and iterating on generated concepts
Artists using reference images to maintain consistency
Content creators making quick edits without switching platforms
Professionals needing to modify specific parts of generated images
These features would make Merlin.ai an even more comprehensive solution for creative professionals and enthusiasts alike.
G
Guy
endu I wonder if we could use images in Projects to build an image from multiple images? E.g., if I upload 3 character images, then use all those characters in a new image.
G
Guy
endu Another thought on this was to be able to change the style of an image. Say you have a realistic photo - it would be great to be able to take the exact same image, but make it a watercolour, or a drawing, etc.
endu
Guy got it :) .will try to implement it
cc: Kunal Verma
G
Guy
endu As part of this, do you have the ability to look at inpainting, too? - To make slight changes to an image that you do like. You paint over an area and use a prompt just to change that one area.
endu
Guy yess we are bringing this to Bonkers/ multi image model generator;
"Use one image as a basis for image generation, to be able to copy things such as style, mood, characters, and so on." - this will be second in line as it will require some more time to implement
"character sheet (multiple images of a character)" - this we havn't planned as of now but we need to look into its execution as as of now the responses generated by models are not good enough .. :/
G
Guy
endu Great! Thanks for looking in to those things!
The character sheet may well be part of the second one, where you can copy a character.
It would be great to be able to get consistent characters in some way.
endu
under review
Merlin
Thank you for your feedback, Guy ! We're glad to hear your thoughts, and we’ll definitely consider it for our roadmap.
G
Guy
Merlin Thanks!
Merlin
Merged in a post:
Consistent characters
G
Guy
Be able to upload a character sheet (multiple images of a character) or single image of a character, to then generate new images using that same character.