Image Generation
Nano Banana · GPT Image
Auto-sorted — image = reference · PDF/TXT/MD = document (up to 14 images · 10 docs).
Up to 14 images · 25MB per file
Tap to put on or take off the board · Seedream 5.0 Pro up to 10
All outputs in this area are generated by generative AI
Video Generation
Seedance · Kling · Veo · MiniMax · Wan · Gemini Omni · Grok
Gemini Omni notes
Write what you want to avoid as a sentence in the prompt itself (e.g. "Do not do X") — a separate negative prompt field is not supported officially.
Generated videos include SynthID watermarking (official) — invisible to viewers but detectable programmatically for provenance verification.
You can add up to 3 reference videos, each up to 3 seconds — the official docs note that any audio in a reference video is ignored.
Images are used as subject/style references (not the starting frame). Pick a role for each image (face, clothing, background, object, lighting, style) and the official <IMAGE_REF_N> tags are added to the prompt automatically.
Edit and restyle your uploaded video via prompt. The video must be 10 seconds or shorter (official), and we recommend 9MB or less (the same applies to videos entered by URL). (Video editing of uploads is restricted for EEA · UK · Switzerland users)
Continue editing the Omni video you just made. In the prompt below, concisely write only what to change (e.g., "make only the car blue, keep the rest").
Upload a video to extend, or pick one you just made with Omni using [Extend scene] on the result card.
Adds a new scene to the end, up to 10s at a time, for a total of 40s (official). A video you upload to extend must be 10 seconds or shorter; that limit does not apply when extending a video made with Omni. To make the transition seamless, some of the final frames of your original video may be edited (official note). If you use timecodes, 0s means the start of the extended part. If the video you upload has someone speaking, extending it with new dialogue is not supported (official — it works if the video is silent or you add no dialogue).
How to use — 1) Upload a background image and it becomes the first scene of the video. 2) With an arrow (straight move) or freeform line (curved path), draw the direction the camera/subject will move in start → end order (color is for distinguishing targets). 3) Undo = only the last drawn line, Clear All = all drawn lines, Remove image = clear the background too and upload a new image. 4) Write the scene description in the prompt box. The drawing itself does not appear in the final video and is used only as motion guidance (official).
이미지=참조 · PDF/TXT/MD=문서 · 영상=참조로 자동 분류돼요.
Camera · Direction (optional)
Saved in your browser and reused for the next generation. Clear it to turn it off.
Image → Video (first·last frame · optional)
On file drop, base64 is sent directly · 30MB limit
All outputs in this area are generated by generative AI
Turn a line of text or a single photo into moving video — aspect and length are yours to pick
Audio Generation
Seed Audio 1.0 · SeedSpeech TTS 2.0 · Gemini 3.8 TTS · OpenAI TTS · MiniMax Speech · Kling SFX · Mureka · Lyria
Auto-sorted — image = reference · PDF/TXT/MD = document · video = sound generation (Kling Video to Audio · 3–20s).
All outputs in this area are generated by generative AI
Narration, sound effects, background music and songs from one sentence — pick the voice and the mood
AI Video Creation Platform
OpenAI
Google
Anthropic
ByteDance
IMAGE GEN
GPT Image 2.5 · Nano Banana 2 · GPT Image 2
RECAST FULL
HFLUX RECAST · Veo 3.1 · Kling 3.0
3D STUDIO
Set Building · Camera Direction
UPSCALE 4K
Upscale · 1080p · 2K · 4K
AI VFX
Roto · Cleanup · Depth · Motion · Previs
SCENE SCRIPT
Finished Script · Scene Breakdown
REWARD
Campaign · Ambassador · Referral
AI outputs may vary in quality·result each attempt and do not guarantee a perfect result, and credits used (deducted) for generation are non-refundable. By charging, you are deemed to agree to Articles 6·9·11 of the Terms.
Media generated this session appears here
Usage status
이번 세션 사용량
AI outputs may vary in quality·result each attempt and do not guarantee a perfect result, and credits used (deducted) for generation are non-refundable. By charging, you are deemed to agree to Articles 6·9·11 of the Terms.
Shown in real time when you generate images · videos · audio · prompts
Charged credits have a usage validity of 1 year from the payment date; the unused balance may be refunded within 5 years from the payment date. For details, please see the Terms of Service.
You create. We reward.
Get rewarded with HFLUX credits when your campaign entry is selected, your idea is adopted, or you invite a friend.
AI isn't perfect. You are.
Create and submit
Pick a video you made with HFLUX and submit it. Submissions are open year-round, and if it's selected for marketing, you earn credits.
HFLUX Ambassador
Creators with a track record of selected work are appointed as ambassadors and receive monthly credits.
Propose an idea
Send us new feature or improvement ideas. If adopted after review, we reward you with credits.
NOTICE
Reward credits are usable only within HFLUX and cannot be transferred, converted to cash, or exchanged for third-party points. They expire 180 days after the grant date and are non-refundable.
When posting sponsored content created with granted credits, please disclose the material connection. (e.g. #Ad #HFLUX)
Prompt
A short brief becomes a rich prompt tuned to your image or video model
Auto-sorted — image = reference · PDF/TXT/MD = document · video = reference video (up to 30 images · 10 docs · 1 video of 1 minute or less).
All outputs in this area are generated by generative AI
Turn one short line into a prompt the model understands — tuned to your category and video model
Prompting basics — the thirteen labeled sections join into one continuous prompt
FLOW
Keywords · Images · Documents · Video → professional FLOW (any genre & length)
Auto-sorted — image = reference · PDF/TXT/MD = document · video = reference (up to 14 images · 30 on Seedance 2.5 · 10 docs · 1 video).
Style signature (optional)
All outputs in this area are generated by generative AI
SCENE
Finished script (.txt/.pdf/.docx)
Images = references (max 14) · Name a file after a character, location or prop and it applies only to those SHOTs (other names = all SHOTs) · PDF/TXT/DOCX = scene script (1 script · or type below) · 1 video = direction/tone reference.
Style signature (optional)
All outputs in this area are generated by generative AI
SCENE
Automatically organize long scripts/features (.txt/.pdf/.docx) into scenes
Or paste text directly
CANVAS
Connect nodes to create images, video, and audio
Storyboard
Industry standard · 3×3 grid · shot design
Character sheet = reference image · Video = a key frame becomes a reference image and its content is read too · PDF/TXT/MD = document (up to 14 images · 1 video · 10 documents)
Up to 14 · 25MB each · applied to all SHOTs
Shot metadata defaults
Auto = applies industry-standard shot composition
All outputs in this area are generated by generative AI
Break your story into SHOTs, one panel each — shot size and angle follow the industry standard
Upscale
Upscale your finished video to higher resolution — as the final step after editing
Upload your finished (edited) video — upscaling is the final step.
Billed per second (5-second minimum). Videos at 31-60fps cost 2x, and 61-120fps cost 4x.
All outputs in this area are generated by generative AI
Files
My Google Drive — view · upload · rename · delete within hflux
Save directly to your Drive
Files are saved to your own Google Drive (not accumulated on our server).
Access scope
Only files you pick yourself and files this app created are accessed — your entire Drive is not visible.
Google Drive
Library
Archive of outputs from all tabs
· Images · audio · prompts · FLOW are auto-saved in the browser and kept after refresh.
· Videos · music can expire — only those saved to your PC via "Save original (local)" on the result cards in the Video · Audio · Upscale tabs are restored (one click to allow the folder).
· After refresh, galleries refill from the full library; FLOW · storyboards restore editing via their "Open" buttons.
· Storyboard sheets are large — delete old ones when they pile up.
Add PC files (images · videos · audio · text) for permanent storage · deleting removes only from the library (PC original kept) · set a "default folder" to reuse it automatically (one click to allow on reconnect).