most people don't realize the impact of truly multimodal ai models that can generate more than just text

these are the real startup killers

don't believe me?

all of the following tasks required a specific workflow so far, now gemini can do them all: convert sketches to line art and colorize them

https://twitter.com/920606426/status/1899977818563633466

it's so over you can't even comprehend it

this is new "research" from bytedance

100% artificial podcasters will soon exist

7 crazy examples: 2/ this is not just making talking heads, they actively react to what the "other person" is saying

AI lets you hug your dead loved ones again 💀

This is only going to get worse

https://twitter.com/ai_for_success/status/1809940241719132633

AniTalker is yet another high-quality talking head generator!

This one can animate talking faces from a single portrait and input audio with naturally flowing movements and diverse outcomes!

9 cool demos + links ⬇️ 2) Realistic portraits driven by an audio file only

Wear-Any-Way is a new framework for virtual try-on that supports users to precisely manipulate the wearing style of garments.

The method enables users to drag sleeves to roll them up, open coats, and control the style of tucks, among other things.

Some cool examples: Manipulation with click

Less than 24 hours ago, OpenAI released the first glimpse of what Sora looks like in the hands of artists.

This is what unchained creativity looks like!

7 beautiful examples: 1) @shykids

High quality AI generated human videos are here!

Champ can generate videos of anyone with a single image and a bit of motion guidance 🤯

Links ⬇️ This doesn't just work with realistic humans, also with stylized characters from old paintings.

3D content is about to explode!

13 AI methods to generate 3D objects that got released this week: 1) SV3D: Stability AI released a new model for high-resolution, image-to-3D reconstruction.

sv3d.github.io/?ref=aiartweek…

Was interested to see how Midjourney v6 handles the Stable Diffusion 3 prompts I shared yesterday.

Who do you think does it better?

Left is SD. Right is MJ.

15 comparisons:

2) Beautiful pixel art of a Wizard with hovering text 'Achievement unlocked: Diffusion models can spell now.

Less than 72 hours ago, Google showcased their new mind-blowing video model called Lumiere 🔥

Here's a quick breakdown of what it can do: Text-to-Video

The foundation for every video model

The possibilities of AI video editing are gonna be nuts!

Just this week, 10 different papers got released, that show us a glimpse of the future.

Let's take a look at what awaits us: 1) MotionCtrl: High controllability of both camera and object motions

https://twitter.com/1546508709513265153/status/1733110597573562405

High quality AI generated human videos are coming!

Animate Anyone can generate videos of anyone with a single image and a bit of pose guidance 🤯

humanaigc.github.io/animate-anyone…
This doesn't just work with realistic humans, also with cartoon/anime characters.

Exactly 44 hours ago, 12 hours before Midjourney released the Style Tuner and 100 hours before OpenAI is gonna blow our collective minds, Runway released some major improvements to their text-to-video model.

19 crazy examples that caught my eye: 1)

https://twitter.com/1527649679227355138/status/1720135193934041447

The Midjourney Style Tuner is here! 🔥

Less than 24 hours ago, MJ's eagerly anticipated style tuning feature got released!

Here is everything you need to know 👇





ℹ️ Info

This is not a fine-tuning solution like Stable Diffusion Dreambooth or LoRA. You can't bring your own dataset of images. This is more like a style tuner for your specific prompts!

Spiral tutorial time 🌀

In this thread I'm going to guide you through the steps to create your own spiral / optical illusion art with Stable Diffusion and ControlNet. Let's jump in 👇 First of all, you need a good enough GPU. If you have a good enough GPU in your home, you can run this locally. You find installation guides for all operating systems on GitHub:

If you only have potato for a computer, you need a cloud solution.github.com/AUTOMATIC1111/…

Felt inspired by the work of @MrUgleh and tried to recreate the effect. Pretty cool.





For anyone wondering how this is achieved 👇

control_v1p_sd15_qrcode_monster with a strength between 0.45 and 0.85 combined with a black&white basic spiral or other pattern is the trick here.

Highres fix with upscale 2 and denoising strength of 0.65 to further strengthen the effect.

A lot of you wondered how I created my Claire Silver "motion" contest submission: LIVE.DIE.REPEAT.

So I shall deliver. Here's a quick breakdown 👇

At the center of LIVE.DIE.REPEAT. is the endless, always forward moving loop.

I knew I wanted to visualize an entire lifetime within the 12s limitation we were given. I had the visuals for the first, second and last shot in my head, so I asked GPT to help me come up with ideas… https://t.co/qSGg2Gcq2Itwitter.com/i/web/status/1…

Right when I started playing with Gen-1 and Gen-2 by RunwayML, a new text-to-video called ZeroScope XL got released. The model by @cerspense is based on ModelScope and is able to generate watermark free videos with a resolution of 1024x576 pixels.

The results can be quiet impressive, but also cursed. The ZeroScope v2 workflow encourages exploration with a smaller model at a scale of 576x320 and then commit to an upscale with the XL model.

Hey, Listen! 🧚‍♂️ I'm once again looking for #aiartweekly cover submissions! Winner gets $100 and all finalists will be mentioned in issue #40 ✨

Theme is #AISurrealism (sponsored by @ew_corpse)

2 images max
10:14 vertical aspect ratio
Unminted
Like+RT+Tag 2

Submit below! 👇 Rulebook:

If you need some inspiration, check out the wonderful #AISurrealism gallery on FND: https://t.co/NNIDt0wbspaiartweekly.com/challenges/35-…
foundation.app/world/ai-surre…

Midjourney v5.2 just dropped. Here's what's new:

- Outpainting is finally here 🔥
- A new Aesthetic System with improved aesthetics and sharper Images, slightly improved coherence and text understanding
- A new "High Variation Mode" which makes all variation jobs much more… https://t.co/s2MQSmALqGtwitter.com/i/web/status/1…
The new “Zoom Out” feature lets us outpaint an image by a factor of 1.5x or 2x and outpaint the borders of non-square images so that they are square again.

There is also the option of zooming out with a custom aspect ratio.

Cover submissions for #aiartweekly #39 are in 🥳

Theme was "censorship" ❌

From 71 pieces I picked 4 finalists for you to vote on. Now it's your turn to pick a winner 🏆 Vote in tweet poll below! 👇





Finalists are @brockwebb @mind_wank @moelucio @newmediapioneer