Feature Requests

Face Swap applicable to people, animals and objects: To be able to generate continuous illustrations where the characters are consistent and their faces do not change, resulting in different results.

I am interested in using the image-to-image feature through an external API. It would be beneficial if this API could be made available externally, allowing for more seamless integration with other applications and workflows.

Nice feature like in Krea AI. So you can go more creative and have more controls on the output. When you can control the significance (strenght)of each image. Also it would br nice to add more upscaler models for more smooth upscaling (resizing) without make image soapy and with lots of unessesary artifacts. Espessialy when working with photorealism,photography (kind of Scunet,DAT or ESRGAN models). And one more thing: it would be realy good if we can edit custom styles,see what images it containes to replace some assets if needed and tweak it more for more accurate styles results.Thanks. Big Ups Recraft team! Love your product!

feature that can change the style (like animated to real image) but keeping all the details like pose or clothing. fine-tune can kind of do it, but it doesn't do the pose or clothing well at times

The current field length both in the desktop project inputs and especially the mobile app is very restrictive. We all use LLMs to prompt our vision and ideas. Ideogram has huge input length and they condense the text as needed . If not unlimited at least double the accepted input size . It is painful to convert the LLM output to fit the text field

When I'm in a project where I am generating a lot of images that are a similar theme and/or style I leave the previous images on the artboard and click new image to run the prompt again. When I do that my prompt remains (which is a good thing) but the model defaults back to v3. Then I have to go through the whole process again to go into choosing the model I am using. Sometimes I forget and I end up wasting credits on a pair of images that are using the default v3 model which isn't the style I want (if I'm using another model). It would be great if the model stayed at the same model we used previously so we didn't have to go clicking and scrolling to get back to the same model every time.

If you select an area to modify a foot in Recraft and use the 'Modify area' feature more than 5 times, the quality degrades with artifacts on the background outside the selected area. This problem gets worse and worse each time you press the Modify button. This is a problem that is not present in open source tools like Forge that support inpainting, so it needs to be taken seriously and fixed. The fix is simple. You can easily avoid degradation by cutting a selection of the image from the original image and then pasting the modified part back into the original image. Please implement this process automatically.

The application is very intuitive and makes it easy to create projects quickly, which is fantastic. However, as the number of projects grows, navigating through them becomes more challenging. To enhance the user experience, introducing a tagging system could be beneficial. This would allow users to categorize and find their projects more easily based on custom-defined tags, making navigation more efficient and personalized. Another idea would be to enable the creation of groups and subgroups of projects, allowing for a hierarchical structure to better organize content. Lastly, a thumbnail mosaic showcasing all the generated images from different projects could be implemented. This would provide a quick and visual way to browse through content at a glance. These features would greatly improve usability, especially for users managing a large number of projects.

Automatic face restoration when generating full-length people.

When something goes wrong with the app, be it upscaling or generating something, it discounts credits although there was no expected outcome. It should reverse the discount credit(s) to its previous state