Moinzek

When each frame acts as a living painting.

Like this? studiotriple.fr Also, I always find Luc Devroye's site useful for many reasons (luc.devroye.org/fonts.html) Last thing, you can explore some Cargo sites, some very unique typefaces could be there by any chance (but it requires some free time to explore): cargo.site/community

Type twitter: I'm looking to find more foundries that are a bit weirder like BB Bureau (bb-bureau.fr) Any tips?

@crnacura @makemepulse As we say it here, in good old french, this studio is "la crème de la crème" 🙂 And time just proves they keep delivering both creative & technical masterpieces every year (highly recommend people to experiment "Kaizen", from 2024).

I'm incredibly thankful to the team at @makemepulse for sharing their story today. If you're interested in the people behind some of the web's most inventive interactive experiences, you'll enjoy our latest Studio Spotlight: tympanus.net/codrops/?p=117…

@icreatelife Creatures are always great, in a way the scope to imagine and/or create them is just infinite. It's my way-to-go when a new model just released. This was one of the 2 videos made during your 114K event, I was in the "Poppy" mood :)

I’ll just be here creating creatures that don’t exist

@icreatelife @x_aivy @Framer_X Thanks!

@moinzek @x_aivy @Framer_X Looks beautiful

Who else is creating with AI this weekend? Creating another animation right now with @x_aivy and @Framer_X

@markzcsd Thanks Mark! End results may vary a lot depending on the models used, I usually like to to combine them (for example GPT Images 2.0 and then subtle variations applied with Nano Banana Pro, in this image, it was Seedream 5.0 Pro only).

@moinzek I like the style of your images ,they're very artistic

"Light is Paint" - Part II.

@TianweiY @reve @arena @Meta @Google Reve 3.0 before winter comes, at this rate 🙂 Higher DPI for the images soon? It would be awesome!

@reve 2.1 is here. It ranks #2 on the @arena leaderboard, with an almost 30-point lead over @Meta's Muse Image and @Google's Nano Banana 2. Compared to Reve 2.0 just a month ago, we significantly scaled up our layout model—in model size, training compute, and our biggest reinforcement learning training run yet. More importantly, I think we've reached an inflection point in scaling our system and proving our layout bet. While we're probably still using only ~1% of the compute behind today's frontier LLMs, we're already seeing incredibly exciting scaling-law behavior. From here, we're going to scale on every front as fast as we can—while continuing to improve the underlying technology and obsess over every design detail. You should also expect a much faster release cadence. As we continue unlocking recursive improvements to our training and research stack, every generation of the system helps us build the next one faster. We're moving from yearly releases to monthly—and eventually every few weeks. This is just the beginning.

Show us your favorite image or video created today It is your time to connect with other artists and to discover what everyone is working on! This is more than a gallery, it's a chance to swap inspiration and celebrate the work happening across our community. Every piece you share helps someone else feel a little more at home here.

Reve 2.1 is here. The world’s best 4K image model just got better. Greater prompt understanding, world knowledge, and stronger foreign-text rendering.

@GlennHasABeard Thanks @GlennHasABeard 🙂 For now, this lil' creature is still work-in-progress but I already have some interesting iterations of it!

@moinzek Very nice, love that little creature.

Share your art, this space is yours! Share, like, comment, tag a friend to pull them in. Posts like these are a great way to connect with others in your niche!

"LIGHT IS PAINT". More still(s) preview, this time with a 21:9 aspect ratio, with Seedream (latest version: 5.0 Pro).

An other test:

First try with the (very) new Seedream 5.0 Pro and the early results so far are seems pushing into the right direction. Here's a little test with a painterly-look for the environment & costume merged with a real and tangible face for the woman and the little creature as well.

JOB OPPORTUNITY Alright fam. I am looking for two artists comfortable with V2V on Seedance for 5 to 6 weeks, starting end of this month / early August. It’s for an hybrid project with a serious animation studio. It’s obviously paid… Studio will handle shooting + animation and idea is to leverage V2V to properly render their blocking with AI. Bonus if you are in Montreal. Can you DM me your references if you are interested? Also cannot promise I will be able to answer every single DM depending on how busy this gets.

@vanschneider I see "Créativité" written in good ol' french in the nav', I'm damn happy!

Finally upgraded my personal website vanschneider.com to Semplice v7 — All running smoothly. Expect some more updates on my site shortly.

From what I'm testing now (mainly for a cinematic look), Seedream is closer to what GPT Images 2.0 is producing, at least now so very (too, perhaps?) early. Muse would be closer to Google's model, Nano Banana 2 (but it's just my own view, could be way more different to someone else). In the end, it's good to have more models to experiment (and these 2 being released less than a day is a great thing), since each one is always great at something an other one just falls slightly behind 🙂

@moinzek Great!! Based on my early observations, Muse seems closer to GPT-Image 2 than to Seedream. Maybe Google is about to wake up soon… just like it did at the end of August last year, to wrap up the summer.

ByteDance is deploying Seedream 5.0 Pro, a multimodal model that advances image generation and precision editing, especially for computer graphics and mockups. lhc.media/en/media/ai-wa…

I'm trying Muse Image at the same time. It has its own strengths however, for now and for the cinematic quality I'm aiming for, Seedream delivers way better results. It doesn't mean Muse is bad at all, it is great on different styles, like fashion-oriented photography or montages, for example.

@moinzek Thanks for sharing! I’ve seen a few interesting things. I’ll test it alongside Meta’s Muse by the end of the week, on a few slightly complex use cases, to better understand where these two new models really stand.