> Today, DALL-E cost 15$ to use it 115 times, but 90% of the time you don't really get good images, or necessarily exactly what you want.

So you're paying $15 to get 11 great images that are what you want? That still seems pretty cheap.

The point is, the cost includes the failed images. If you wanted the cost to only be for the good images, and you get free bad images, they'd simply charge 10x the amount.

> So you're paying $15 to get 11 great images that are what you want? That still seems pretty cheap

It isn't, stable diffusion + anything v3 or protogen is free

Midjourney costs 10usd/month for iirc nearly 2 hours of gpu time and is better than dalle

Dall-e 2 is completely overpriced and almost obsolete since MidJourney v4.

Hey, I'm getting into using these images more and more. I'm a bit gunshy of all the work that goes into midjourney. Do you know of any good tutorials on how to use it?

I kinda learned while playing with the technology.

Looking at what others are doing is very helpful.

This prompt book has also been the best reference for me, even though it’s not specific to MidJourney: https://openart.ai/promptbook

Awesome book! Thank you!

in truth midjourney does much better on smaller prompts than stable diffusion or dall-e which tend to do better with longer prompts, but my best results have come from image to image along with prompts. The secret superpower of midjourney is that you can use multiple images as image prompts (the rest can use one), first image becomes core structure and composition, second image becomes primary style, third is secondary style etc, but all influence the image. My best results have been from 2-3 images.

other tips: add on the end --ar 3:2 for wide image or --ar 2:3 for tall

most important part of prompt toward front, usually what kind of medium the art should be (coloring book page, stained glass, watercolor, etc), then describe the content and use photographer terms like portrait, landscape, or long shot to describe framing and what's in the image, things like background and foreground help too.

if you know much art or photography theory, it will give you superpowers over normal users.

take some time to look at other people's prompts on lexica and midjourney showcase, if you want to play with stable diffusion for free first I recommend playgroundai.com which only requires a google account for about 300-800 free spins a day (they throttle you a bit after 300), but I think it's a good way for people to dip their toes in the water, but midjourney does tend to give better default results

Btw you can use it for cheap using this GUI: https://dall-e.sonnet.io/

Why does this make it cheaper? Couldn't this site just be stealing people's API keys?

> Why does this make it cheaper?

API calls are ca. 7 times cheaper than credits. (up to 1000 images vs. 150 when paying with credits)

> Couldn't this site just be stealing people's API keys?

Thanks for asking. Feel free check the source and host it yourself: https://github.com/paprikka/dall-e-ui (takes 2 min with Vercel)

Context: I made this GUI for myself when I was comparing SD and Dall-e for a small personal project. Some people seemed to like it, so I shared it.

https://twitter.com/rafalpast/status/1591138659297726464?s=2...

regarding dall-e, eventually we would up with stable diffusion, which lets you tinker to your heart’s content at no cost on fairly cheap hardware.

my hope is that an open source alternative to chatgpt will emerge and see similar iterative, community developed models for specific purposes

A Netflix-style model would be a lot nicer, indeed.

Actually, they are laying out their model quite clear in the sign-up form:

- Always available (no blackout windows)

- Fast responses from ChatGPT (i.e. no throttling)

- As many messages as you need (at least 2X regular daily limit)

So I guess this is basically the Netflix style you are asking for.

https://docs.google.com/forms/d/e/1FAIpQLSfCVqahRmA5OxQXbRln...

Thanks! Sounds reasonable.

The downside of all-you-can-eat is that it doesn't really work for occasional use.

Wouldn’t that require a different approach to paying for computation? From OpenAI perspective there is no difference in spent resources whether you are still tinkering or know exactly what to generate in DALL-E. Which seems fair.

DALL-E is massively overpriced for the compute it uses. The current DALL-E price is based on what they think they can get away with, not on compute costs.

That's... How pricing works

There's a lot more to product pricing that compute costs...

Yes, but it is in their interest to monetize this. Many people I talked with are fully willing to pay for it after it goes behind a paywall. OpenAI improved the user experience of their Playground with ChatGPT and not so many people taking out the wallet for DALL-E 2. It is fair to mention that the free alternatives of DALL-E 2 are equally capable, if not better.

I've found that combining the two AI's gives me more value out of DALL-E 2 i.e. using ChatGPT to generate prompts for DALL-E 2

I'll pay up to $150/month for chatGPT but it's a massive nope if they have a credits system.

I'll even consider paying more but if the access is metered, my usage pattern will change in precisely the way you describe. At that point, it's not worth it.

How often do you use it to justify paying 150 a month?

Every day, mostly as a coding assistant. Been a HUGE timesaver.

>15$ for 200 message

The GPT3 api is way cheaper than that, so I would expect the final price of chatgpt premium to be quite low.

If it was 15$ for 5000 messages would you still worry so much about wasting credit?