Film Creator Online Using AI Models

A 12-scene short film needs final renders in 90 minutes

You open Flixly with 240 credits left and a script for a 90-second chase sequence. The first move is to pick the right models for each stage rather than guessing.

Define the sequence and credit budget

List every shot on paper first. Scene one requires a static establishing shot in 4K, scenes two through seven need motion from reference frames, and the last five need lip sync on a 22-second monologue. Allocate 20 credits per 5-second clip for Seedance 2.0 and 15 credits for Kling 3.0 background plates. This leaves 80 credits for testing and final polish.

Generate base frames with GPT-Image 2.0

Start at the AI Image Generator. Prompt "rain-slick alley at night, 35mm lens, 8K" and set aspect 16:9. GPT-Image 2.0 returns four 1024x576 options in eight seconds. Pick frame three, upscale it once inside AI Image Tools to 2048x1152, then save it as base01.png. Repeat for the next six keyframes. Each upscale costs 3 credits.

Verify frame consistency

Open the same prompt again but add "same character as base01.png, red jacket" to force continuity. GPT-Image 2.0 keeps the jacket color across outputs when the seed stays under 12345.

Add motion with Seedance 2.0 and Kling 3.0

Move to Image to Video. Upload base01.png, select Seedance 2.0, set duration to 5 seconds at 24 fps and camera move "slow push in". The job finishes in 42 seconds and produces a 120-frame MP4. Switch to Kling 3.0 for the wide establishing shot because it handles rain particles better at 30 fps. Both files land in your dashboard library at 1080p.

Reference additional plates

For the running sequence use Reference to Video. Drop three stills of the alley plus a 3-second phone clip of the actor. Set motion strength to 0.7. The output matches the actor silhouette while adding rain streaks that were missing from the reference.

Handle dialogue and lip sync

Record the 22-second monologue on your phone. Head to Lip Sync Video, upload the 5-second chase clip from earlier, paste the transcript, and choose Gemini 3.1 Flash TTS voice "Alex". The model aligns mouth shapes to phonemes in one pass and returns a 1080p file with embedded audio. Check timing at the 14-second mark where the word "left" lands exactly on the head turn.

Stitch and caption in one dashboard view

Download the six clips. Use Shorts Generator to concat them in order with 0.5-second crossfades. Add auto-generated captions at 48-point size, burn them in at the bottom third. Export the final 1080p 90-second file. Total credits used: 187.

Model comparison table

Final verification steps

Play the exported file at 1.5x speed to catch any frame drops. Check audio levels peak at -6 dB. Open the file properties to confirm 23.976 fps and AAC audio. The 90-second film is now ready for upload.

You now hold a finished 1080p short film created entirely inside Flixly. Repeat the same workflow on your next project by starting at the text to video page.