I thought editing an image in ChatGPT meant giving it another prompt and hoping it would change the right thing without messing up something I already liked. Change this, leave that alone, don’t touch the background. Rinse and repeat if I didn’t like the results.
What I somehow missed was that clicking the finished image opens a whole editing workspace. From there, I could use different tools that would let me erase or change the image in some way. Once I tried the tools, I found they made editing much easier than I expected, though there’s one limitation that can get annoying fast.
I can change one thing without rebuilding the whole image
Comment lets me make the change I want without rewriting the whole prompt. I only saw the options because I clicked the image once ChatGPT finished creating it. That’s when I saw options like Comment, Remove BG, Erase, and Resize. To try Comment for the first time, I clicked the pillow behind the cat and told it to turn the pillow blue. By doing this, I avoided having to tell ChatGPT not to touch anything else in the image.
I wanted to see how well it handled multiple comments, so I told it to add a coffee mug, make the cat smile more, and turn the frame in the back into one with a nature picture.
I also uploaded an image of my dog to see how well it handled my images, and it did a surprisingly good job. I first had ChatGPT change her red sweater to blue. Then I clicked on the edited image and saw the Comment options at the top. I had it turn her scrunchie blue, and it did. I was so ready to complain about the edit, but had nothing to say at the end.
I can remove what I don’t want instead of rebuilding the image
Erase targets one spot, while Remove BG needs a clear subject
Erase is easy to use. I only have to select the person standing on the left side of my garden photo and click the removal option. ChatGPT did a good job filling in the scene once the person was removed. The path, grass, and flowers still looked natural enough afterward, which is really the whole point.
Shrink the brush when the object you want to erase is close to something you want to keep. A smaller brush gives you more control and helps avoid removing nearby details.
Remove BG worked differently. I tried it on that same garden shot first, but the image had so many trees, paths, and flowers that there wasn’t an obvious subject to isolate. It wasn’t useless, but I didn’t get the edit I was looking for. ChatGPT did a great job of removing the background in the Labrador’s picture. It removed the park, leaving me with a transparent cutout of my dog. With the result I got, I could easily reuse her on another background instead of just being stuck with the original photo.
I didn’t have to remake the image just because the shape was wrong
I pushed the same pet selfie from 9:16 to 16:9, and it still held together
Just like the other tools, Resize is easy to use. When I click on the Resize option at the top, I can choose from options such as:
- Square 1:1
- Portrait 3:4
- Story 9:16
- Landscape 4:3
- Widescreen 16:9
I first changed the pet selfie to 9:16, fearing the bunny would be cut off or the dog would end up awkwardly squeezed into the frame, but ChatGPT kept all three animals there and just gave the scene more room vertically.
Then I tried the pet selfie in 16:9 on the same image and ran into no issues here either. It widened the scene, kept the cat, dog, and bunny together, and didn’t make the image look like someone had just hacked off the top and bottom. That’s really why I’d use Resize. If I already like the image, I don’t want to make it again just because I need a different format.
The free tier is where this starts to feel cramped
Image creation and uploads come with tighter limits
You can still create images and edit them if you’re on the free tier. However, it does come with its limitations. OpenAI says image creation and uploads can have stricter limits than on the paid plans. That matters when you’re working on something important and need to create and edit more images than you initially intended.
I rarely stop after creating and editing one image. There is always something I want to change, or I want to see if ChatGPT can polish it even more. All that back-and-forth eats up my usage limits, and I eventually hit the limit. With the Free tier, users can hit those limits sooner, which may mean having to pause and wait before continuing the project.
If Free keeps getting in the way, Go is worth a try
It can give you more room for images without jumping straight to Plus
If paying for ChatGPT isn’t an option, then Free is what it is. You can be more selective about which images you upload and which edits are worth another try, rather than burning through the limit on every little change.
But if you keep getting cut off halfway through something, you can try ChatGPT Go before switching to Plus. It costs $8 a month in the U.S., and OpenAI says it gives you 10 times more file uploads and image creation than Free. That could be enough if your main problem is running out of room while you’re still working on an image.
I’m done treating every image change like a new prompt
What changed for me is pretty simple. If ChatGPT gives me an image I mostly like, I don’t automatically start over anymore. I can click it, fix the part that bothers me, remove something, or change the format without rebuilding the whole thing from scratch. Remember that you can review your saved images under Images, instead of trying to find the old chat again. So if ChatGPT gets an image almost right, don’t immediately write another giant prompt. Instead, click the image first and see what you can fix from there.