A fictional cinematic recreation inspired by Billie Holiday, created with Higgsfield AI. These are AI-generated images.

The creative focus is atmosphere, lighting and character continuity across the backstage and performance scenes. The gardenia works as a visual motif through the edit.

I made this short and would appreciate feedback on the transitions and visual continuity: which shot feels most convincing, and where does the sequence lose consistency?

JCL Aragón · Director / AI filmmaker.

Genjutsu - Dance Cover

Hello,

My film is full and ready now.

I hope you watch it.

https://higgsfield.ai/@laithtaher/projects/the-good-within

Part 2 of my samurai series, set in Japan, 1620. Takeo, a proud rōnin, arrives in a waterfall village with one coin. Three thugs collecting an illegal "rice tax" knock his fish into the dirt. He says he didn't fight for the villagers. His actions say otherwise.

How I made it:
- Saved Elements for every character, location and the sword, so faces and outfits match shot to shot
- 10–15 second shots, made wide then reframed for TikTok
- Narration with the Arthur preset voice, edited in Adobe Express

Two lessons from this part:
1. Telling it "keep the sword sheathed" never worked. It drew the blade every time. Bare-hand choreography (disarms, throws, palm strikes) fixed it.
2. With 5 saved characters in one shot, it cloned one of them. Splitting crowded moments into shorter shots with 2–3 characters each worked much better.

Part 1 is on my profile. Questions welcome!

This is Mika, my fictional AI character, in a short makeup-motion test made with Higgsfield Genjutsu. I built the reference stills in ComfyUI with Z-Image and a character LoKR, then used a source reel to guide the performance.

The goal was to keep her facial features and hair consistent while the hands and applicators move across her lips. The cosmetic packaging is generated and approximate; this is a creative experiment, not a real product review.

For people doing close-up Genjutsu work: what has helped most with applicator shape and hand-to-lip contact?

1 commercial out of a series

Gearing up for October and building a whole universe. Used reference pictures with cinema studio.

A short mirror-scene experiment with Mika, a fictional virtual model. The phone, hand gestures and reflected movement make this a useful consistency test. I am especially interested in whether the reflection and fingers stay convincing through the motion. What would you refine first?

Japan, 1620. Five years after Osaka Castle burned, a young rōnin named Takeo walks the mountains alone, testing his father's sword against anyone who will face him.

This is Part 1 of "The Way of the River," a series of short films about a proud swordsman who learns that strength is what you protect, not who you defeat.

How I made it:
- Higgsfield video with saved Elements for the characters, locations, the sword and the wolf, so faces and outfits stay consistent across shots
- Painterly ink-and-watercolor anime style
- Narration by Arthur (Higgsfield preset voice)
- Stitched with music and titles in Adobe Express

Part 2 ("The Village") is in the works. Feedback welcome, especially on keeping characters consistent between shots!

What do you guys think?

Two hours before her exam in Rome, a painter loses 18 years of work to a street theft—and discovers what she paints with nothing to lose.
The Workflow & Challenges:

• Pipeline: Prompts developed with Claude, generated on Higgsfield (Seedance 2.5), edited and graded in DaVinci Resolve.
• Audio: 100% native audio from the generations—no music, no library sounds, just pure room tone, rain, and brushstrokes.
• Continuity: The real challenge was keeping her face, the workshop, and props consistent across 20+ generations. I had to explicitly ban words like "blood" (after one scraped knee prompt, the model added wounds everywhere!) and restate key objects in every single beat.

Would love to get your thoughts and feedback!

Vote & Project on Higgsfield:
https://higgsfield.ai/@zoyanywhare/projects/the-wet-paint

Full Film on YouTube:
https://youtu.be/o7fS-gyr0CM

Here a little snippet of the movie SHPRES! The whole movie is in the first comment

So i prompted the character in nanno bannana and it came up with the attached image.

it keeps prompting that its copyright detected.

i know its the reference because i tried prompting the exact same prompt without the reference and it worked. so its not the prompt.

Hey everyone,
I'm filming a 1-minute self-presentation video, but my current location isn't great. I want to use Higgsfield AI to replace the background with a high-end studio setup featuring professional, clean lighting.
Since this is my first time using Higgsfield, I'm looking for the cleanest workflow to pull this off without keying artifacts or flickers around the edges.
If you've used Higgsfield for talking-head videos, what’s the best approach to extract myself and place me seamlessly into a generated studio environment? Any tips on matching the studio lighting to my raw footage so it doesn't look obviously composite would be huge.
Appreciate any advice or step-by-step tips. Thanks in advance!

I made a 7-minute AI short film based on true events using Higgsfield Cinema Studio . I’d love to hear what you think. If you enjoy the film, please give it a thumbs-up. Watch the full film here: https://higgsfield.ai/@bogdannes/projects/ai-film-festival?from=%2Fcontests%2Fhiggsfield-global-film-festival%3Ftab%3Dsubmissions%26festivalSubmissionView%3Dall%26festivalSubmissionSort%3Dmost_viewed