john_noumenonic
No user record in our sample, but john_noumenonic has activity below (stories or comments). Likely we have partial data — the full bulk-load will fill profiles in.
No user record in our sample, but john_noumenonic has activity below (stories or comments). Likely we have partial data — the full bulk-load will fill profiles in.
In the case of pictures taken with your phone, the "true" answer would be "real" as it's not a synthetically generated image, just some post-processing/cleaning-up. The overall classification is going to be affected by…
Wasn't aware of that context in the anime art community, that's good to know about, so thanks. Currently, artist drawn anime images weren't a primary stylistic focus, so the gap makes sense.
It shouldn't classify the example you explained as AI generated, but we are looking at expanding some functionality for similar use cases to that. W.r.t AI-upscaling, the current model isn't looking for it specifically,…
Do you know what base model was used to generate the image? Was it SD v1.5?
We're generally seeing an accuracy of 90% on test sets, with its distribution primarily being images generated by diffusion models and the "real" images that are more ground level imagery. We'll have to take…
True enough, though we're interested in stress testing our model to see where the gaps are right now.
That's great to know actually, especially the low-quality/distorted images bit.
Are you able to use the image upload box to select a file or drag-and-drop images?
The preliminary model gets ~93% accuracy on an "in-the-wild" test dataset of ~10k images pulled from the widely used models, i.e. midjourney v5/v6, SDXL, DallE-3, etc.