Bug Reports

When you select an image and brush "strokes" from the brush tool the merge object is disable but you still can merge using the shortcut or right click mouse button option menu

When generating images, the number of images to be generated is set to 2, yet more and more images are generated without permission while they are being generated, and they are reduced by about 15 credits every time at once, so why is this?

In Spanish: The seamless models are not working well, they fail far from the edges. I understand that you are generating an image that is not seamless and you are moving horizontally and vertically about 256px, and the error is therefore at that distance from the edges. Check out these images created with both raster and vector seamless models. These sometimes inconspicuous lines of color can end up in a final product if not thoroughly reviewed. I told you about it on Discord in January, but this is still the same, it's still failing. Is there a plan to fix it? For this function we are currently using midjourney with the --tile parameter. But I would like to see Recraft in the fight for this type of function doing it perfectly, since other things you are doing great.

Hello, it seems that currently text_layout is supported only in images/generations endpoint, would be nice to have this parameter supported in other endpoints as well

I have found an issue with the images/imageToImage endpoint. It does not appear to take into account controls when they are passed in and therefore my colors are not being applied. We have followed the documentation here - https://www.recraft.ai/docs#generate-image If we add the formData controls argument and then add the colors list the colors are not being applied. We have also tried the txt2Img endpoint and this applies colors fine but this takes in json input not formData. This is an example we have tried in NodeJS to attach colors using the img2img endpoint: formData.append('controls', JSON.stringify({ colors: [ ...(inputColors?.map(color => ({ rgb: [color.r, color.g, color.b] })) || []), ], no_text: noText, }));

When I click "create new project" the download icon appears and it keeps endlessly downloading with no results. It happened after I've bought additional 400 credits for my basic account. Since then the "create new project" is not responding even though I've updated my account to Advanced. The mobile app works properly.

When creating an image I often get a "server is too busy" error. So I don't get any image creation. However, this still counts against by 50 credit quota, and I see my credits reduced even for trying.... Though no images are produced. This is BS! Please fix this.

I have a paid account. I keep getting an error that says 'Oops try again later'. I'm paying to use the app and no one is answering my emails or fixing the problem.

I'm working on a prompt. I have run it with various styles including some of my styles. It was working fine at first I ran it 4-5 times, then the same prompt started failing to execute with an error message that it may not be for all audiences. That's nonsense. There is nothing in the prompt that would create anything but an image with a G rating. Once I got this warning/error message I kept trying the same prompt. I changed the artistic level and the style. Some of these runs have worked, most have failed with the same error message. If your AI is 'thinking' of creating a prompt with anything but a G rating then you need to look at how it's being trained and what data sets it being trained on. If the AI can't understand and correctly interpret this prompt but instead interprets it as explicit or violent then that presents a problem because not knowing what words or phrases it's incorrectly interpreting, what's triggering it, I'm stuck in a time wasting guessing game trying to get past a censoring AI that isn't smart enough to correctly interpret a prompt. I'll post this then send via email the link to the project so you can see for yourself the prompt and the images I have generated with it so far.

When we generate an image with text when we switch to variate the image shows some kind of the "code" "placeholders" of the text.