Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Midjourney web experience is now open to everyone

midjourney.com

Post reply on HN
Print view View original

101–110 of 182 posts

Re: Midjourney web experience is now open to everyone

#101

Post by geor9e »

SoftMachine wrote: post #25

Open to everyone, but you must have Google or Discord account. Has the definition of everyone changed?

Touché. But it's certainly leagues more "everyone" today than it was yesterday. It used to require arcane knowledge of deep Discord server bot navigation. I gave up after 20 minutes and never figured it out. Today I tried it in seconds.

I have about 40+ google accounts for reasons, so I don't begin to understand the aversion some have to registering a burner google/discord/facebook/etc account under their hampster's name, but many of my closest friends are just like you so I respect it anyway, whatever principle it is.

Re: Midjourney web experience is now open to everyone

#103

Post by ethanpil »

smusamashah wrote: post #38

Ideogram 2.0 was also released today, and it's nerfed (anatomy is lot worse than 1.0 now) just like StableDiffusion versions after 1.5 which is very disappointing. Well good thing we have Flux out in the open now, both midjourney releasing web version or ideogram releasing there 2.0 on the same day after 2 weeks of flux won't redeem them as much. Flux Dev is amazing, check what SD community is doing with it on https:…

Why are they getting worse?

Re: Midjourney web experience is now open to everyone

#104

Post by demadog »

pizzathyme wrote: post #3

This is a interesting company to watch in the Gen AI space since they don't have all the same restrictions as the bigger companies. Crazy this took them so long, and also crazy that they got so far through a very confusing Discord experience.

The only restrictions they have are that you can't imagine President Xi, nor can you mention anything PG-13 like bathing suits.

Re: Midjourney web experience is now open to everyone

#105

Post by quesomaster9000 »

vunderba wrote: post #100
quesomaster9000 wrote: post #79

And it still can't draw a spork yet [1] 1: https://canaidrawasporkyet.org/

MJ can't do it, but took me 10 seconds genning locally on an SDXL checkpoint: https://gondolaprime.pw/pictures/sporks-generative.jpg It's a rather arbitrary metric. All it proves is that the token "spork" wasn't in midjourney's training data. A FAR better test for adherence is how well a model does when a detailed description for an untrained concept is provided as the prompt.

Yup it's an arbitrary metric, but I tried cajoling various image models into generating spork pictures with highly detailed descriptions (I have ComfyUI & AUTOMATIC1111 locally, and many models), which lead to me creating the site.

I'd say a better test for adherence is how well a model does when the detailed description falls in between two very well known concepts - it's kinda like those pictures from the 1500s of exotic animals seen by explorers drawn by people using only the field notes after a long voyage back.

Re: Midjourney web experience is now open to everyone

#106

Post by vunderba »

quesomaster9000 wrote: post #105
vunderba wrote: post #100

Earlier quoted context omitted.

MJ can't do it, but took me 10 seconds genning locally on an SDXL checkpoint: https://gondolaprime.pw/pictures/sporks-generative.jpg It's a rather arbitrary metric. All it proves is that the token "spork" wasn't in midjourney's training data. A FAR better test for adherence is how well a model does when a detailed description for an untrained concept is provided as the prompt.

Yup it's an arbitrary metric, but I tried cajoling various image models into generating spork pictures with highly detailed descriptions (I have ComfyUI & AUTOMATIC1111 locally, and many models), which lead to me creating the site. I'd say a better test for adherence is how well a model does when the detailed description falls in between two very well known concepts - it's kinda like those pictures from the 1500s of…

Here's a test against the full Flux Dev model where we try to describe the physical appearance of a spork without referencing it by name.

Prompt: A hybrid kitchen utensil that looks like a spoon with small tine prongs of a fork made of metal, realistic photo

https://gondolaprime.pw/pictures/flux-spork-like.jpg

The combination of T5 / clip coupled with a much larger model means there's less need to rely on custom LoRAs for unfamiliar concepts which is awesome.

EDIT: If you've got the GPU for it, I'd recommend downloading a copy of the latest version of the SD-WEBUI Forge repo along with the DEV checkpoint of Flux (not schnell). It's super impressive and I get an iteration speed of roughly 15 seconds per 1024x1024 image.

Re: Midjourney web experience is now open to everyone

#107

Post by demadog »

thot_experiment wrote: post #63
seatac76 wrote: post #36

Earlier quoted context omitted.

Try creating something mocking Xi Jinping.

wait really? i'm not gonna sign up for midjourney but it's surprising to me that this would be restricted

Yep - they self-censor themselves according to China's whims, just so they can have China access now only to be banned by the great firewall 1 year from now after Chines startups scrape all their image outputs for training.

Re: Midjourney web experience is now open to everyone

#108

Post by rchaud »

seatac76 wrote: post #36
pizzathyme wrote: post #3

This is a interesting company to watch in the Gen AI space since they don't have all the same restrictions as the bigger companies. Crazy this took them so long, and also crazy that they got so far through a very confusing Discord experience.

Try creating something mocking Xi Jinping.

Why is AI needed for that? Political satire doesn't require anything more than adding a text balloon to an already existing picture.

Re: Midjourney web experience is now open to everyone

#109

Post by vunderba »

quesomaster9000 wrote: post #105
vunderba wrote: post #100

Earlier quoted context omitted.

MJ can't do it, but took me 10 seconds genning locally on an SDXL checkpoint: https://gondolaprime.pw/pictures/sporks-generative.jpg It's a rather arbitrary metric. All it proves is that the token "spork" wasn't in midjourney's training data. A FAR better test for adherence is how well a model does when a detailed description for an untrained concept is provided as the prompt.

Yup it's an arbitrary metric, but I tried cajoling various image models into generating spork pictures with highly detailed descriptions (I have ComfyUI & AUTOMATIC1111 locally, and many models), which lead to me creating the site. I'd say a better test for adherence is how well a model does when the detailed description falls in between two very well known concepts - it's kinda like those pictures from the 1500s of…

Welll.... there's a hundred ways we could measure prompt adherence everything from:

- Descriptive -> describing a difficult concept that is most certainly NOT in the training data

- Hybrids -> fusions of familiar concepts

- Platonic overrides -> this is my phrase for attempting to see how well you can OVERRIDE very emphasized training data. For example, a zebra with horizontal stripes.

etc. etc.

Re: Midjourney web experience is now open to everyone

#110

Post by evilduck »

liuliu wrote: post #92
evilduck wrote: post #80

Earlier quoted context omitted.

Just to clarify for other readers, Draw Things has support and provides download links to quants but no iOS device can run the full precision model which means you will get slightly different and usually lower quality output than stuff you may see elsewhere, even if you use the same settings. It's still damn impressive though.

Quality issue should be mainly due to using FP16 accmulators for GEMM in M1-M2, A14-A16 devices (it is not a problem for SD v1 / SDXL models due to smaller channel count). This is changed to FP32 accmulators for GEMM in these devices with 1.20240820.1 release. q8p should have comparable quality to non-quantized models (in Draw Things, it is called FLUX.1 [dev] (Exact)).

Hey the creator! Thanks for all your hard work. Put in a tip feature so I can give you money.
Post reply on HN