1. Love the Capcut 3D zoom! Sadly I need to grab my iPad just to get those effects. But wanna try using it on something, I have an idea already.

  2. All told I burned through about 30/40 credits doing this: not bad for $5. But if you had more patience and a few more dollars, I reckon you could get even cooler results!

  3. "A photo of a confused grizzly bear in calculus class"

  4. This is weird and it's also weird the same prompt caused it to fail for someone else!

  5. I thought realistic faces weren't possible, that they disabled the possibility. Is there newer news?


  7. this is really weird, the results shouldn鈥檛 be this all over the place. Something isn鈥檛 right here!

  8. I feel framing this whole thing in this way would be the best way to push against all of this bullshit, I mean who would want their services to be called a gamble?, I also believe we indeed gotta keep pushing against this, people from OpenAI is bound to know the outrage is going around the community and we gotta keep on drilling into their heads we don't want this... at the end I don't mind paying, just make it reasonable, I would be willing to pay say 30 bucks a month for a service that would give me 50 or so prompts per day as long as they upped the generated context up to 6 or 10 again...either way they kinda digged themselves into this issue due to having given to us such an amazing experience just to take it away.

  9. I don't think that's plausible. The pricing isn't just made up, it's based around what it costs to run the servers - it's a literal physical cost in electricity and hardware.

  10. how can you be sure of that?, that is the point of a company isn't? to profit?, with the lack of transparency they got going they very well could be artificially inflating the prices.

  11. I can't be 100% sure, but it would just be very unusual for a startup like OpenAI to prioritise profit over user growth at this point! Their big message right now is onboarding a million users in the next month or two.

  12. It is obvious that the racial diversity in the generations of people has broken the whole system. Now the generation core does not express the most approximate and high-quality result, but looks for the average for all aspects of generation, and therefore the results become blurry. This and the ridiculous paid system killed the whole idea of 鈥嬧媎all-e, lol, but everything started so well.

  13. I鈥檝e been trying MidJourney recently (I鈥檓 an early DALL路E tester but my eye is starting wander) and I鈥檝e been very impressed.

  14. I'm... concerned about Shrek's cleavage.

  15. I know it may not work... Could you try Hornio?

  16. I wish they let you generate small images (225px) for free, and pay for downloading/generating the big version (4k or so).

  17. Yeah, but apparently it's making the little images that's the hardest bit computationally so it doesn't actually make that much of a difference their side.

  18. That's a really good idea! I want to test those sorts of terms with a bit more rigour, somehow - the reason being, it's harder for me to judge 'by eye' how much an output successfully captures 'the look' of an entire community website! (vs 'does this look kind of like a Pixar character'?)

  19. In theory, no (OpenAI stripped out all NSFW images from the training data so it literally doesn't know what certain... things... look like)

  20. Quite so. Which I totally get in terms of uploading a photo of someone, it's just frustrating when it's a photo that DALL路E generated 30 seconds earlier!

Leave a Reply

Your email address will not be published. Required fields are marked *

Author: admin