91–100 of 137 posts
After getting access to the beta, combined with all these HN posts -- I've determined DallE2 is neat but no where as great as the initial samples made me believe.
Worth noting that DALL-E automatically “rejects attempts to create the likeness of any public figures, including celebrities". So, you wouldn't be able to get an image that included the 4 Liverpudlians. It does allow you to create fake faces. Might be fun to try and recreate Miles Davis Tutu, Aladdin Sane, Piano Man.
Earlier quoted context omitted.
OK, so I give you license to use this URL I just generated to generate your own stuff. It's pay for action (send me a penny if you find anything worthwhile), and the copyright is owned by the payee: https://images.google.com/
The URL has not been "generated" in the same sense. You are retrieving an existing string. The images from google are not "generated" in the same sense, they are indexed from google's search algorithm. The generative models, specifically for DALLE here, compute pixel unique images. You might say these models index a subset of an extremely high dimensional space (pixel count * RGB color values) using a query. Traditio…
I wonder how long the novelty of DALL-E will persist. HN seems to upvote anything titled "I did X with DALL-E". This is a fun post, but it's not that interesting or surprising. Still worth a look don't get me wrong, but personally didn't learn anything new from it. (eg recreating the famous pink Floyd cover with "Outline of prism on a black background in the middle of scene splits a beam of light coming from the left…
"DALL-E" in titles is about to be replaced with "Stable Diffusion." The beta website is already live but the interesting part will be specialized fine-tuned models based on public weights. There should be more technical experimentation since Stable Diffusion weights are only a few GBs and inference can be run on any recent GPU. There might also be more controversies because it can create more uncensored images. Someb…
Although AI artists will destroy a lot of jobs, it will also create demand for new jobs for people who specialize in “paint overs” – taking a high concept output created by AI artists and touching it up to perfection. Or perhaps even beyond just a paint over, and into the realm of recreating an entire AI artwork but with a human touch to get details just right. Looking forward to it.
Here's another spin on this: Imagine you're a software developer. In the near future, your manager wants a feature implemented in your company's app. He throws together a short mail with requirements and sends it to the prompt engineering department. The prompt engineers fix a few typos, clean up the grammar and pepper a few secret sauce keywords around like "in the style of firefox", "in the style of kde". This get…
DALL-E is still highly probabilistic in its judgement. For instance, in this article, it keeps putting "fire" in the the background on something that is likely to be on fire, rather than lighting up the person. I have a similar experience. In my own experiment, I can't get DALL-E to turn off the street lamp at a bus stop in the darkness. I've tried "no light", "broken street lamp", etc.; no use. Any mention of "stree…
I have experienced violent or harmful settings to be avoided by DALL-E. E.g. setting a person on fire. Same with drowning - seems to be impossible/hard to generate
Have anyone given a prompt to Dall-e of designing a company website and included “make it pop!”? Maybe the AI will finally get what designers always complained about annoying clients.
I suggest learning how to use a few graphics packages instead when you have a need. eg, this evening my wife wanted to print a birthday card for a dog that she is boarding. I fired up Inkscape and discovered a bug wrt handling an imported photo, scaling it and cropping it with a mask. lol! I was able to export it to a .png and print from something else.
Anyway, I was a little unfortunate and IT is still rather shakey. However, discussions about DALL-E are way more interesting than actually using it yourself. You see some seriously intelligent solutions to getting a desired image that is surely a form of programming. DALL-E does not even have a real natural language pre-processor because it is hamstrung to dump pejorative terms etc. This means that I can draw the classic, dreadful plan of the cargo hold of a slaver ship from the 17/18C and its cargo but I doubt DALL-E can.
The q&a about this thing are way more complex and interesting than its actual output! A is right, I it isn't. DALL-E is quite interesting but it really isn't intelligent. The intelligent bit is getting the input right for the desired output. Perhaps another model could be developed for that.
Earlier quoted context omitted.
"DALL-E" in titles is about to be replaced with "Stable Diffusion." The beta website is already live but the interesting part will be specialized fine-tuned models based on public weights. There should be more technical experimentation since Stable Diffusion weights are only a few GBs and inference can be run on any recent GPU. There might also be more controversies because it can create more uncensored images. Someb…
I dunno, I don't see a lot of examples on that page where Stable Diffusion outperformed DALL-E 2.