promptcrafted

Promptcrafted is an AI design and image creation studio run by artists. You can’t prompt without a human, and you can’t design without an artist.

[Upscale Conf @upscaleconf]

🖼 How is AI revolutionizing the creation process?

Sofía López (@sofialomart, @freepik), Emilio Nicolás (@emailnicolas, @Magnific_AI) and AI artist Araminta K (@araminta_k) explore this with @LinusEkenstam moderating the #UpscaleConf panel on AI Image Generation

[Araminta @araminta_k]

I've been playing around with some more mystically themed Glifs.

Today I made a pretty comprehensive Horoscope Glif. It allows you to privately put in a natal chart, or just provide your basic signs, and receive a horoscope that is tailored to today's specific astrology.

(Link)

[Dreaming Tulpa 🥓👑 @dreamingtulpa]

UniHair can create 3D hair models from single-view portraits, handling both braided and un-braided styles!

It uses a large dataset and advanced techniques to accurately capture complex hairstyles and generalize well to real images.

Links ⬇️

[GLIF @heyglif]

New Tutorial Drop!

Today we're recreating this recipe app by @ostrisai , but adding a stylistic twist! Full video in the thread.

[Magnific.ai @Magnific_AI]

🔥 Mystic v2 is out! 🔥

Magnific is no longer just an upscaler, it’s now one of the top AI image generators out there, creating images up to 4K quality.

This isn’t just an update; it’s a fresh start for us.

[Javi Lopez ⛩️ @javilopen]

IT'S FINALLY HERE!

🔥 Mystic v2 in Magnific 🔥

“Any sufficiently advanced technology is indistinguishable from magic.” — Arthur C. Clarke ✨ Mystic v2 is the most advanced AI generator to date that can now output up to 4k resolution images!

Info 🧵👇

[21 @twentyone21___]

EXPLORERS

[Dreaming Tulpa 🥓👑 @dreamingtulpa]

InstantDrag can edit images quickly using drag instructions without needing masks or text prompts!

It learns motion dynamics with a two-network system, allowing for real-time, photo-realistic editing.

Links ⬇️

[Matteo Spinelli @cubiq]

I gave an LLM a few comfyui nodes and the basic relationship between them. The LLM was then able to generate an upscaler with tiling, controlnet and Florence2 captioning. Working with json files is very easy for llms and with training it could become a pretty powerful tool

[Jake Dahn @jakedahn]

we got another one!

I'd like to introduce my next flux dev LoRA, it's called flux-latentpop.

It's great for t-shirt designs, screenprint-style posters, character portraits, and pretty much everything in between.

[fofr @fofrAI]

low res video footage lora anyone?

[fofr @fofrAI]

I made a glif that gives you an image that's the opposite of what you asked for.

It uses llama-405b on a high temperature setting and flux-dev for images.

https://t.co/MTuzqgcPeF

[Araminta @araminta_k]

**New LoRA Drop**

I haven't felt super inspired to release the models I've been working on until today - Really happy with this Polaroid stylized model. I'm not sure the deal with the name "polaroid" so I just called it "Pola Style" to be safe. :)

I've really been enjoying doing this series of vintage film styles.

Linking the links below. Thanks to @replicate for gifting the training gpus to train this bad boy!

[Jake Dahn @jakedahn]

Ladies and gentlemen, I think we have a banger on our hands!

This is, without doubt, the best LoRA I have ever trained 🤗

link below!

[Justine Moore @venturetwins]

Help I’m genuinely invested in the Gordon Ramsey AI cooking series

[fabian @fabianstelzer]

this whole space benefits tremendously from creators like Mint - not a good look for RemixAI to just pull non-commercial models without attribution, esp when you're explicitly told not to, that's not how open source works or flourishes

the model is the medium is the message

[Araminta @araminta_k]

I experienced a situation with a startup that left a really bad taste in my mouth - it was a mess.

I felt it would be important to share - to highlight behavior that I, and other, creators have experienced.

@getremixai reached out to me on Thursday of last week to let me know they were featuring some of my open source flux LoRAs in their next big announcement the *next day*. (First I heard of it.)

It was an odd conversation, and it shouldn't have been. I have conversations like this all the time with brands.

I kept asking which models they were highlighting, and it wasn't until I'd asked at least three times that they finally sent me a list.

A couple of my models - particularly Enna Sketch which is created with my personal illustrations - are not fully open source. I pointed this out, and their representative argued that actually they had a free tier with access to the LoRAs, so it was *technically* not commercial. (thats not how that works...)

I asked them to be more mindful and double check the rest of the licenses on the models and thought that was the end of it.

Bizarrely, it was NOT. More in thread.

[fabian @fabianstelzer]

made with the alpha of the incredible new Glif Flux Lora trainer, courtesy of the amazing work by @ostrisai @araminta_k @rvorias

i trained a lora on 15 weird medieval illustrations, so now you can prompt whatever your heart desires in weird medieval style

run it here: https://t.co/TyLZnggerB