The hardest part is getting the camera to behave how I wanted, especially to do handheld doc-style operating. The model just did not want to do it. Also, it will put in music to some shots no matter how hard you try to get it not to.
Everything with the rats and the otters is my favorite, but what really got me to push that scene were the initial test shots. In fact, the opening 60 seconds of that scene, where you see the otters fishing and the rat walk up out of the fog, are basically two 30 second multishot chunks, and they are the first takes, I believe. I was stunned by how good the lighting and characters looked, and how good the "performances" were, especially from the rat. The model nailed the polite creepiness that creates such a sense of unease.
I really wanted the dad otter to grab the blade of the rat's knife and hold on for dear life, just like Caytlin Stark in the first episode of GoT. I thought it would be a good mirror to the mom otter grabbing the boiling pot of water. But it just refused to look good...
TALES OF GRIMWICKE β Chapter I is 22 minutes of AI fantasy animation inspired by the many anthropomorphic animal stories I read as a kid, with a dark twist:
The model: Seedance 2.5 (@ByteDanceSeed_ ) via @Segmind_ai. 847 takes generated, mostly 30-second multi-shot prompts, over 5 hours of footage. 127 clips made the cut. About $4.5k in generation, four weeks from first take to publish.
Characters and locations came out of @midjourney. Using MJ for the location plates was a revelation: the cinematography and art direction are a level above, and it renders far better as a plate than Nano Banana or GPT Image. Nano Banana (@GoogleDeepMind run on @fal) did the heavy lifting editing those plates, building anchor frames and turning MJ characters into turnaround sheets. Then I let Seedance do as much of the framing and camera operating as possible. That gave the best results.
A lot of what I do in my Unreal Engine films came with me: a handheld, imperfect camera operator instead of smooth moves, focus hunting, and above all weather and atmospherics. The model does water, snow, smoke and rain really well, and leaning into that is what keeps it from looking like a Temu Pixar film.
The tools I built: a portal (Node on @flydotio) that holds the bible, the refs, every prompt and every take, with a Claude agent (@claudeai) that writes the prompts to the house rules. A Premiere Pro (@AdobeVideo) panel that browses the archive, drops selects on the timeline, versions the project, strips music, and sends fixes back to Seedance. Everything is done in 720p with no upscaling, because it looks terrible. Narration is my own voice through @ElevenLabs speech-to-speech. Music from @Artlist_io.
None of this would exist without @PJaccetturo and @Kavanthekid. Their narrative work, Nexus and Chronicles of Bone, proved long-form AI film was possible, and their podcasts, newsletters and X posts, updated constantly as the workflow changes under everyone's feet, showed how to actually do it.
Chapter II is currently in the works and I will release it initially on my YouTube channel, The Darkest Age (@CharlieDriscollAI). Consider subscribing there to see it as soon as it drops!
There were only a handful that I actually submitted back to Seedance for editing, maybe five or six clips. These were mostly to do character swapping where the model messed up and put in the wrong character but I liked everything else about the take. I want to explore editing more
@maanvis81@ByteDanceSeed_@Segmind_ai All voices are done in Seedance 2.5, with prompt directing. The narrator's voice is my own that I changed with elevenlabs
@doublezero2fe@ByteDanceSeed_@Segmind_ai Good question, there are a handful of full 30 second chunks, maybe 3 or 4 or so, where I used every cut. Definitely a few 30 second oners. But often I was combining 30 second multishots to get what I needed
@TheTimeTraveler@ByteDanceSeed_@Segmind_ai A lot of almost incredible takes had to be tossed because of some small defect that made it I possible to cut in, or just didn't fit in some way.
@ZePatriote_2@ByteDanceSeed_@Segmind_ai This is 100% Seedance 2.5 with no 3D integration, mostly Midjourney location plates and character sheet references which initially started in MJ then completed with Nano Banana
Thanks, glad you enjoyed it. Agreed. It was hearing what other artists do to get to their final results that helped reframe everything for me. Once you realize it costs a few thousand dollars and can take dozens of tries on certain gens, it helps you prepare going into a longer form project like this.
@dxiaolong@ByteDanceSeed_@Segmind_ai As of now it's still prompt adherence for more complicated scenes. I tend to try to push it to do bigger battles and more intricate blocking, which takes a lot of iteration to get right
@Uvacha_ai@ByteDanceSeed_@Segmind_ai Thanks! I am hopeful we will have Seedance 2.5 quality on our own hardware like H3 within a year, at the rate things are advancing.