4 ms·
Oof 800 by 800 kills a lot of use cases
by knollimar 2mo ago
Oof 800 by 800 kills a lot of use cases
- asdfsa32 2mo agoflash vs fine details. Pick one.
- Doohickey-d 2mo agoGemini "flash" models have an option for media resolution, including a high resolution option for screenshots.
- skeledrew 2mo agoAt what price point?
- wongarsu 2mo agoFor most use cases you can fix that in the harness. Just give the model a tool to request a crop of specific coordinates of any image it has in its context. Call the tool "zoom" and it should be intuitive for the model Maybe there are some use cases where you need high detail everywhere at once, but for OCR of small text and the like a zoom ability should be sufficient
- embedding-shape 2mo agoFor really dumb models I've also had success automatically cropping it into a grid of N images with the max size, then processing each cell individually, then once all been processed, do one final call with resized image + all other context previously generated per cell. Basically a workaround to the image dimension restrictions without loosing fidelity. Works well with even dumb 7B models. Can't remember if I stole this idea from some existing public harness though, can't remember. If someone knows of public harnesses that do this already, please share them :)
- dotancohen 2mo agoDoes this not loose context? Especially e.g. in fonts where the character pairs 0O 1I 1l Il may be difficult to differentiate?
- skeledrew 2mo agoThat's what the grid crop should handle. The detail is retained at that level, and then everything is logically stitched together again using the lower-res-full-image as reference. That's going to be 2x token usage at minimum though.
- Aeroi 2mo agoi think the claude cookbook has a file that does this, called tiling.
- johndough 2mo agoMight still be fine. The most recent crop of vLLMs proactively use whichever programs are available on the system (e.g. ImageMagick or PIL) to "zoom in" by cropping subimages if they can't quite make out the details.
- knollimar 2mo agoDownsizing a higher res image to lower res means the zoom will be blurry.
- andai 2mo agoThey process the original image file with Python on the local device. (And I've seen the web chats do this with their "computer use" features too.) The really wild one is even blind models will do this and they'll try to run stats on the pixels to figure out what it looks like... the even wilder thing is that it kind of works!
- knollimar 2mo agoIf the API accepts only 800 by 800, the aegument youre making is "fix it in the harness". I don't think the n by n subgrid fixes this the way most harnesses do, as it'll fail to count things if you have more overlap and fail relatiomships if you have less
- tjoff 2mo agoThat seems weirdly specific? And if you are counting things it should be trivial to note the position of your items and not double-count them, no?
- adastra22 2mo agoThey’re not talking about zooming, hence the quotes.
- knollimar 2mo ago
- shadyr 2mo agoIt might also be due to its experimental status. Wouldn't surprise me if the GA version allows for larger input. Either that or the eventual pro version.
- Chnmy 2mo agowhat are these use cases?
- knollimar 2mo agoAnything where there are symbols representing in space (e.g. schematics). Thats pretty broad
- stronglikedan 2mo agoI don't know about a lot. Probably more like a few. I take a lot of screenshots for various reasons, and over 800 seems like I could have done a better job framing and cropping.
- hedgehog 2mo agoMulti-turn, expose a tool that extracts crops. Works great.