Open your workspace
Sign in with the account supplied by your administrator. New accounts are currently created by an admin during the private pilot.
STUDIO KNOWLEDGE BASELearn Studio.
Learn Studio.
Make the idea visible.
Create through conversation: images, videos, music or talking avatars. Add references, choose supported settings, review credits and continue from results in the same saved project. Industry templates and reusable prompts are available after sign-in.
FEATURED FILMAfter darkFind a look for your brand
GETTING STARTED
Your first creation, step by step.
The workspace has navigation and projects on the left, chat in the centre and Generated results on the right. Drag the dividers to resize the panels. Start with an idea in everyday language.
Choose what to create
Choose Image, Video or Music beside the composer, or select Avatar creation under Creative task. Choose a model. Open Options for supported ratios and resolutions; use the duration selector for video or music. Avatar duration comes from the speech recording. Available controls change with the model.
Tell the assistant
Describe your goal, audience, subject and visual style. Use the microphone if you prefer to speak. The assistant may ask for missing details.
Add references
Open the + menu to upload images, documents, video or audio. Choose each image’s role from its thumbnail menu: Product, Character, Scene, Environment or Style. For an avatar, attach one portrait and select uploaded speech in Options. Each model has its own reference limits.
Review and create
Check the captured brief, prompt and credit estimate. Once it is ready, click Generate or type Generate now or Generate a video based on this brief. The chat checks access, credits and the available generation budget, then queues one result or explains what is blocking it. Admins can generate during the pilot; other users can prepare and save briefs until generation is enabled.
Find your results
Keep chatting to create another version or new content. On a completed image, choose Make video to bring it into the same chat as a reference, or Use in chat to create another image. Describe the motion or change, choose a model and duration, review the credit cost, then generate. Results appear on the right and in Assets. The project saves the conversation, draft, settings and references; open it from the sidebar to resume the chat.
Make a good prompt in one sentence
Name the subject and action, setting, composition or camera movement, light and mood, then say what should stay faithful to your reference. For a video, describe distinct shots and transitions.
Research a website in chat
Paste a public HTTPS website link: “Research https://example.com and suggest three ad angles.” The agent reads up to two linked pages and shows source links. Then ask it to turn the findings into a poster or video brief in the same saved chat.
For research without a link, the agent can search the web. Check sources and approve factual claims before publishing. Login-protected, paywalled and pages requiring JavaScript may not be readable; paste text or upload a document instead.
Research has a daily allowance of 20 requests per user, or 50 for administrators, and also counts toward the normal chat allowance. Attempts count even if a page cannot be read. Media generation still requires review and credits.
REFERENCE TO VIDEO
Bring your subjects into motion.
Upload up to six images with the + control, label each as a character, product, scene, environment or style, then choose a reference-capable video model.
Start with value
Grok Imagine References offers 480p and 720p with native audio. H3 Max References offers 480P, 768P and 1080P; references are fitted into square images for predictable input costs. Both support 5 to 15 seconds. Review the credit estimate before generating.
Explore other looks
Vidu Q1 References creates a fixed 5-second clip. Wan 3.0 and Seedance support multiple subject references too. Seedance 2.5 is a premium option. Describe which details must stay consistent; results vary.
Frames versus references
H3 Turbo uses two images as start and end frames. For several independent subjects or styles, select H3 Max References or another model marked with six references. Changing the model changes the available duration, ratio and resolution.
MARKETING WITH STUDIOStart with a task.
Start with a task.
Build the creative through chat.
Start with the prominent Creative task selector above the studio chat. TikTok product story, Cinematic commercial, Social product poster and UGC product video provide editable starters and task-specific starting settings. Choose Quick tour for a guided introduction to the controls. The language switcher offers English, Bahasa Melayu and Chinese for the main workspace controls, starter prompts and agent replies. You can also type /marketing to see task options. Supply product facts and references, answer the assistant, choose a direction, then review the prompt and credit estimate before generating one image or video.
Custom creative task
A personalized workflow shaped around your goal through chat.
TikTok product story
A thumb-stopping hook, a visible demonstration and one clear action. Built for vertical viewing, without promising virality.
Cinematic commercial
Build a miniature brand film with timed shots, deliberate lighting, sound and a memorable finish.
Social product poster
A fashion-forward product hero with a deliberate layout, approved headline and mobile-readable CTA.
Poster ad design
Design a finished ad with a product hero, expressive typography and approved copy. Start from an industry poster or describe your look.
Avatar creation
Upload one clear portrait and speech audio using +. The audio sets the talking video duration.
Campaign angles
Share the product, audience and approved benefits. Compare three ad angles before choosing one.
Customer feedback ad
Paste genuine reviews or customer comments. Turn their needs into an ad direction.
Headline variations
Compare three headlines while keeping the same product, layout and visual style.
Video hook variations
Compare three opening hooks, then create one video with the chosen opening.
Localize an ad
Paste approved copy and name the language and market. Keep offers and claims accurate.
Product photoshoot
Upload a product image and choose a studio, lifestyle or dramatic hero direction.
UGC product video
Plan a natural hook, product demonstration and CTA in your chosen video duration.
Cinematic storyboard · 1×1 · 1 shots
One storyboard image with 1 cinematic panels. Plan shot size, camera angle, action and cuts. Choose 2K or 4K when supported for clearer panels.
Cinematic storyboard · 2×2 · 4 shots
One storyboard image with 4 cinematic panels. Plan shot size, camera angle, action and cuts. Choose 2K or 4K when supported for clearer panels.
Cinematic storyboard · 3×3 · 9 shots
One storyboard image with 9 cinematic panels. Plan shot size, camera angle, action and cuts. Choose 2K or 4K when supported for clearer panels.
Cinematic storyboard · 4×4 · 16 shots
One storyboard image with 16 cinematic panels. Plan shot size, camera angle, action and cuts. Choose 2K or 4K when supported for clearer panels.
One direction, one reviewed generation
Headline, hook and campaign tasks offer alternatives for you to select. They do not automatically generate a batch. Paste approved facts, genuine reviews and the source copy for localization; Studio does not browse competitors or measure campaign performance. Marketing tasks need the creative AI agent enabled by your administrator. Review generated text, product details and claims before publishing.
POSTER AD DESIGN
Design the artwork through conversation.
Select Poster ad design under Creative task, or ask the chat to design a poster. Start from an industry poster sample and adapt it to your business.
1. Build the brief
Share your product, audience, brand colours, language and campaign goal. Upload the product or logo with +. Supply your approved headline and CTA, or ask the assistant to suggest copy for your approval.
2. Choose the art direction
Compare visual treatments such as expressive editorial typography, playful product collage or a premium product hero. The assistant specifies composition, palette, graphic details, text hierarchy and margins for your selected ratio.
3. Create and refine
Approve the direction and copy, choose an image model and review credits before generating. Inspect the actual spelling, logo and packaging, then request focused changes in the same chat. Results are generated images; editable design layers and print-ready exports are not provided.
CREATIVE SHOWCASE
Imagine this for your brand.
Find a style you love, then create your own product ads, posters and videos inside Studio.
Video stories
Explore real generated examples across 16 SME industries and cinematic styles. Use these examples to find the direction for your next campaign.
Image concepts
Explore commercial posters and product imagery across SME industries.
USEFUL TO KNOW
Better results start with a clearer brief.
Can I create a similar style for my brand?
Choose a template inside the workspace, upload your product reference and describe the result you want. The assistant helps adapt the direction. Results vary by model, settings and references.
When should I upload a reference?
Attach a reference when the result should follow a real product, person, setting or style. For a product ad, a clear product photo helps preserve shape and packaging.
How do credits work?
The cost changes with model, resolution, video or music duration, and speech length for avatars. Review the estimate in the workspace before creating. During the private pilot, generation is enabled for admins.
READY TO CREATE?
Sign in to Studio ↗Start with one idea.
Choose a style inside the workspace or describe something entirely your own.
VIDEO TO PROMPT · ADMIN PILOT
Turn a reference video into a creative brief.
Open Video to prompt beneath the chat, upload a short clip or paste a public YouTube, TikTok or direct video link. Choose a video model and duration, then Analyze video.
1. Add a short video
Use a clip up to 30 seconds. File uploads support MP4, MOV, WebM and MPEG up to 8 MB. If a social site blocks link access, upload the clip instead.
2. Review the breakdown
The agent describes the scene, camera movements, style, sound and timed cuts, then prepares an editable recreation prompt. Analysis uses the service budget; it does not generate media.
3. Continue in chat
Choose Use prompt in this chat. Ask for changes or replace the product using your image references. Review the generation cost before generating.
A starting point for your own creation
The prompt approximates the video; it cannot recover its original prompt or guarantee an identical result. Uploaded video is used for analysis only. Attach image references separately when the selected generation model needs them. Analysis is currently available to admins.
TALKING AVATARS
Turn a portrait and speech into a talking video.
Select Avatar creation in the studio Creative task menu. Use + to upload one clear portrait and a speech recording, then review the cost and generate in the same chat.
1. Prepare the presenter
Use one well-lit face image. Choose Character on its thumbnail. To change clothing or the background, prepare the portrait in Image mode first, then use it for the avatar.
2. Add the voice
Upload 5 to 120 seconds of speech, under 8 MB. You can record it or create speech in Voice Studio and upload the downloaded file. Options lets you select the lip-sync audio when several recordings are attached. Speech controls the dialogue and video length.
3. Choose quality and cost
MiniMax H3 Max Lip Sync supports 480P, 768P, 1080P and 2K. FlashTalk offers lower-cost automatic portrait or landscape framing at 448/768 dimensions. Choose H3 Max Lip Sync when you need ratio or resolution controls. Kling Avatar v2 offers Standard and Pro choices for people, animals and stylized characters. OmniHuman 1.5 offers audio-driven expressions and movement: use audio shorter than 60 seconds at 720p or 30 seconds at 1080p. Review the measured audio length and credit estimate before generating. Review the audio duration and estimated credits before generating. Avatar generation requires an eligible account, sufficient credits and an available service budget.
Use your supplied files
These models animate the supplied portrait to the original uploaded speech. Text in chat does not replace the audio or alter the image. Use portraits and voices you have permission to use.
MUSIC THROUGH CHAT
Give your content a soundtrack.
Choose Music beside Image and Video in the studio composer. Describe the mood, genre, instruments and purpose, then refine the direction with the assistant.
1. Choose your sound
MiniMax Music 3 supports instrumentals and original songs with lyrics, up to five minutes. Stable Audio 2.5 offers instrumental backing tracks and sound design, up to three minutes. Use Options to choose Instrumental or Song with lyrics.
2. Review before creating
Choose a model and duration. Review the music brief, lyrics and credit estimate, then choose Generate music or ask the chat to generate it. Duration is an upper limit; tracks may finish earlier.
3. Listen and continue
Your track appears in Generated results and the Music library. Play it, download it, or ask for another version in the same chat. Downloaded tracks can be used in your editing workflow; adding music to a generated video is a separate editing step.
Availability
Music generation follows your account permissions, available credits and service budget. Generation is available to eligible accounts with sufficient credits and an available service budget.
REFERENCES IN CHAT
Add a reference with one +.
Open the + menu to upload images, a written brief, a short video or music. Images appear as thumbnails; choose their purpose from each thumbnail: product, character, scene, environment or style. Add a detail to preserve, or remove the reference with ×.
Bring context into the conversation
Documents support TXT, Markdown, text PDFs and DOCX under 2 MB. Video references support MP4, MOV and WebM up to 30 seconds; music references support MP3, WAV, M4A, AAC, OGG and FLAC up to two minutes. Media files must be under 8 MB. For image and general video tasks, video and music are analyzed when you send your message, then their creative summary guides the brief. In Avatar creation, speech audio goes directly into lip-sync generation without this analysis step. Media analysis is currently admin-only and uses the service budget; generation is a separate action. Uploads alone do not trigger analysis.