ComfyStudio Pro

I don't really know what I've made here.

@DeadSunStudios Looks great!

Generated this with LTX audio-to-video on my PC. The 2nd clip is upscaled to 4k using Astra 2 from Topaz Labs. Curious if it's a noticeable difference after uploading due to the compression. I've been posting all my stuff in 1080P.

@peterom I come and watch this almost everyday. Brings me joy and laughter!

@ComfyUI I've been waiting for this! Amazing!

With the recent updates, in addition to real human video generation with Seedance 2.0, you can now also use an AI character for video creation. Upload custom characters and create a consistent identity with cinematic control using Seedance 2.0. Try it now: links.comfy.org/4n6tUC3

@Jenny_MommaLion Thank you!

@comfystudiopro Wonderful!

ComfyStudio v0.1.11 is live. New in this release: Director Mode music-video passes, Flow AI, source proxies, commit renders, export fixes and MORE! To show what Director mode can do in practice, I made this music video for AI 90's indie band Ashtray Turnpike entirely in ComfyStudio: "Falling Apart for You". Nano Banana for the initial images. Local LTX 2.3 for video. Also attached are images of my timeline and new Flow AI node based workspace. Yes in ComfyStudio, you can edit, timewarp, add effects like camera shake, film grain, glows, etc. Animations are keyframe-able giving you incredible flexibility once your AI generation is done. Video tutorial coming soon. Get ComfyStudio for free on the links below. All of this built on top of @ComfyUI check this out! @yoland_yan comfystudiopro.com github.com/JaimeIsMe/comf… can @LTXStudio do all this?

@DeadSunStudios Hah exactly! Weezer was one of the bands I referenced on sumo for this song! Just typical soft loud indie music from the 90s.

@comfystudiopro Trying to figure out which band this reminds me of. Kinda Weezer or Jimmy Eats World?

@DeadSunStudios Awesome! You should check out the full length video I made with LTX 2.3. Posted it a couple days ago.

Old LTX audio-to-video lip sync test I never posted. 2nd half it really fell apart. The cool part to me is taking a Suno song and putting a face with the voice. Then it seems real. Goth rock chick taking me back to the early 2000s Evanescence. Going to cover all the genres.

@ingi_erlingsson @ostrisai hah I love this

first time training an LTX LoRA and I think I've summoned a demon - pretty sure it's not supposed to sound like this. @ostrisai am I doing it right? 😅

@cfryant @AIWarper I came here to say this! It also feels to me that's what's going on.

@AIWarper It's the focal length. Changes a person's appearance even though it's the exact same subject.

Bouncing back and forth between GPT-Image-2 and Nano Banana Pro is very hard to retain the exact likeness or am I just crazy? Do these two people seem like the same person? Is it just the lighting that is throwing me off?

WAN 2.2 feels way more natural to me for motion/animation. The movement just has a better flow. But LTX 2.3 has real advantages: - audio conditioning / lip sync - up to 15 sec shots in ComfyStudio - ~2x faster generations on my setup WAN 2.2 is amazing for natural motion, B-roll, and detail shots. LTX 2.3 is still the practical choice when I need speed, longer clips, or lip sync. Both have a place but for this test WAN 2.2 is the clear winner in my opinion.

Did a quick i2v test between LTX2.3 and WAN2.2 on my 5090. LTX 2.3 is at the top. Wan 2.2 at the bottom. Same image references and prompts! Which do you like better visually? I know which one I prefer!

@el_mejnun @MedOumoumad Ya this is a great idea! I'd like to learn as well.

@MedOumoumad just wanted to thank you for your contributions on oss video model loras. as image generation backgrounded people like me your lora ideas are like an oasis in the desert. I would really appreciate video tutorials about how do you decide, dataset curation, training parameters etc.

@tarumainfo man seedance is insane. Youre insane too. Great work!

If you could walk through any door right now, where would it lead?

@comfystudiopro Thanks for bringing this to our attention. 🤝

@ComfyUI not sure if this is happening to everyone but whenever I buy credits, I'm landing on the cool 404 page. You can see the URL says /payment/success. And I've got my credits. It's just weird to be taken to the 404 page.

@AIWarper Love this! I can see a whole music video like this working. She moves room to room in her apartment dancing.

Here’s another one. This time generated completely for free with LTX 2.3 Just gave it audio + a start frame with a prompt of her vibing 💃

@ComfyUI Will do! How about if I just add a pro at the end? ComfyStudio Pro? Hah open to suggestions...

@comfystudiopro We are excited!!! Yeah please 🤞😂

@MedOumoumad @noamiKenKorem Kicking ass like always! This is great!

SDR to HDR from ComfyUI, I've trained a LoRA over Qwen Edit 2011 based on the principle used in hdr-lumivid.github.io A research I've had the previlige to work on with @noamiKenKorem for video HDR on LTX. Links for the LoRA and workflow (+ my fun grading node) below.