Diving deep into stable diffusion
MULTIMODAL AI
-
GPT-4 and DALL-E 3 in 2023: AI will not dominate
By
–
If you are impressed by #chatGPT or #Dalle, wait until 2023, GPT-4, DALL-E 3, etc. Other models from other companies (for example Google with public Lambda). But please in advance: NO, AI will not dominate the world, and I don't know what nonsense I hear each time.
-
CDN Integration Makes DALL-E Image Generation Simple
By
–
Clever integration making DALL-E dead-simple to use — a CDN which lets you generate new images just by requesting an appropriate URL:
-
Z-depth in synthetic scenes transforms 3D artist workflows
By
–
Using z-depth with synthetic scenes will be a boon for 3D artists to do quick grey boxing or kit bashing, and quickly emulating a render. You have immaculate depth in 3D tools already, why bother with monocular depth estimation? Previously covered here: https://
creativetechnologydigest.substack.com/p/stable-diffu
sion-20-out-adding-a
… -
Reasoning in Visual Perception: Distinguishing Squares from Circles
By
–
The former sounds perhaps stranger, so here's an example. Let's say you have to tell whether a given image contains a square or a circle — a canonical perception problem. Sounds easy enough if you have a well-trained visual system, right? How would reasoning come into play?
-
Midjourney Spider-Man Image Generation Progress April December 2022
By
–
.
@Midjourney generating images about Spider-Man in April 2022 vs December 2022. The progress in 8 months is simply breathtaking! Via @ParkerMolloy #AIart #AIArtCommuity #AIArtwork -
AI-Generated Content Indistinguishable From Reality With Fine-Tuning
By
–
Sure but that's only cause I barely tried to give it a specific style. Now imagine this with a bit more effort, finetuned to a specific style. It'll be impossible to distinguish from real. Whether that's good or bad I'm not to judge
-
AI Accelerates Frozen Section Interpretation in Surgery
By
–
During surgery, frozen sections of tissue are obtained for key decision making (e.g. resecting a tumor) but there are problems w/ speed & accuracy of interpretation. Using #AI, a potential rapid fix for this age old problem https://
nature.com/articles/s4155
1-022-00952-9
… @natBME @AI4Pathology @BWHPath