Gemini 2.5 Flash Image

This is the gpt 4 moment for image editing models. Nano banana aka gemini 2.5 flash is insanely good. It made a 171 elo point jump in lmarena!

Just search nano banana on Twitter to see the crazy results. An example. https://x.com/D_studioproject/status/1958019251178267111

I digitised our family photos but a lot of them were damaged (shifted colours, spills, fingerprints on film, spots) that are difficult to correct for so many images. I've been waiting for image gen to catch up enough to be able to repair them all in bulk without changing details, especially faces. This looks very good at restoring images without altering details or adding them where they are missing, so it might finally be time.

Half the time I ask Gemini to generate some image it claims it doesn't have the capability. And in general I've felt it's so hard to actually use the features Google announce? Like, a third of them is in one product, some in another which I can't use, and no idea what or where I should pay to get access. So confusing.

Very impressive.

I have to say while I'm deeply impressed by these text to image models, there's a part of me that's also wary of their impact. Just look at the comments beneath the average Facebook post.

I don't get the hype. Tested it with the same prompts I used with Midjourney, and the results are worse than in Midjourney a year ago. What am I missing?

That lamp example is pretty impressive (though it's hard to know how cherry-picked it is). The lamp is plugged in, it's lighting the things in the scene, it's casting shadows.

FYI, this is the famed nano-banana model which has been now renamed to gemini-2.5-flash-image-preview in LMArena.

All these image models are time vampires and need to be looked at with very suspicious eyes. Try to make a room - that's easy, now try to make multiple views of the same room - next to impossible. If one is intending to use these image models for anything that requires consistency of imagery, forget it.

I love that it's substantially faster than ChatGPT's image generation. It takes ages, so slow that the app tells you to not wait and sends you notification when the generation finishes.

I've had a task in mind for a while now that I've wanted to do with this latest crop of very capable instruction-following image editors.

Without going into detail, basically the task boils down to, "generate exactly image 1, but replace object A with the object depicted in image 2."

Where image 2 is some front-facing generic version, ideally I want the model to place this object perfectly in the scene, replacing the existing object, that I have identified ideally exactly by being able to specify its position, but otherwise by just being able to describe very well what to do.

For models that can't accept multiple images, I've tried a variation where I put a blue box around the object that I want to replace, and paste the object that I want it to put there at the bottom of the image on its own.

I've tried some older models, and ChatGPT, also qwen-image last week, and just now, this one. They all fail at it. To be fair, this model got pretty damn close, it replaced the wrong object in the scene, but it was close to the right position, and the object was perfectly oriented and lit. But it was wrong. (Using the bounding box method.. it should have been able to identify exactly what I wanted to do. Instead it removed the bounding box and replaced a different object in a different but close-by position.)

Are there any models that have been specifically trained to be able to infill or replace specific locations in an image with reference to an example image? Or is this just like a really esoteric task?

So far all the in-filling models I've found are only based on text inputs.

“Internal server error

Sorry, there seems to be an error. Please try again soon.”

Never thought I would ever see this on a google owned websites!

If this can do character consistency, that's huge. Just make it do the same for video...