Sogni / Blogs & Newsletters

Storyboards that move. One chat, prompt to video.

Storyboards used to be the slowest part of any video project, sketch the hero, lay out the beats, hand the sheet to an animator and pray nothing drifts between frames. Sogni Chat collapses all of that into a single conversation.

You describe a character, GPT Image 2 renders a clean reference, then a storyboard sheet with timecodes and on-screen captions. Then you hand the sheet back to the agent and Seedance 2.0 turns it into a finished social video, all inside chat.sogni.ai.

The walkthrough below follows one real chat session: a cyberpunk anime heroine living through her morning self-care routine, shown here as one storyboard example that becomes a finished clip.

The finished Seedance 2.0 clip, surfaced right up front before the walkthrough begins.

Want to make a video like this? Follow the step-by-step walkthrough below and this exact flow takes you from the first prompt to the final render in Step 09.

The welcome screen is already a workflow.

Head to chat.sogni.ai and the home screen greets you with a pinned Featured workflow — GPT Image 2 + Seedance 2.0 panel: build rich storyboard keyframes, then render a finished social video from the same visual plan.

Three demo chats sit underneath it so you can scroll, replay, or fork the finished result before you start your own: Storyboard Video Demo, GPT Image 2 Demo, and Character Sheet Demo.

Sogni Creative Agent 1.0 welcome screen with the pinned Featured workflow panel: GPT Image 2 + Seedance 2.0, plus three demo cards for Storyboard Video, GPT Image 2 and Character Sheet.
The home of chat.sogni.ai. The Featured workflow tile is your express lane from prompt to finished video.

Sign in to claim your monthly credits.

Hit Sign In and the Sogni mascot pops up to wave you through. Every account ships with 400 free Spark credits a month so you can take Sogni Chat for a real spin and see how the agent feels before you ever pay for anything.

No account yet? The Sign up link at the bottom of the modal takes a username and email, and you are creating in under a minute.

Worth knowing. Free monthly Spark goes toward Sogni's broader open-source model lineup. Premium models like GPT Image 2 and Seedance 2.0 are pay-as-you-go with Premium Spark, no subscription required. If you want to stay on the open-model side, newer options like LTX-2.3 and Wan 2.2 are there too.

Welcome to Sogni sign-in modal with the Sogni mascot, username and password fields, Remember me checkbox, Sign In button, and 400 Free Credits Monthly badge.
One sign-in unlocks every tool: image, video, music, storyboards, and the Creative Agent itself.

One tap on the featured tile, you are on the express lane.

Once you are signed in, the greeting flips to "Happy creating!" and the Featured workflow tile is still front and centre. Click Start storyboard video and the agent pre-loads the right prompt shell, the right model chain, and the right confirmation flow for the GPT Image 2 + Seedance 2.0 pipeline.

Prefer to learn from a finished example first? Open one of the three demo chats underneath and the entire conversation, prompts, reasoning, panels, and final video, replays in the timeline.

Signed-in Sogni Chat home with the greeting 'Happy creating!' and the pinned Featured workflow tile for GPT Image 2 + Seedance 2.0.
Same tile, signed-in greeting. Start storyboard video kicks off the whole pipeline.

Two dials in the top bar. Pick once, the rest is automatic.

The top-right Default Media Quality menu picks the routing tier for images and video. Pick once, the chat carries it across every tool call in the session.

Media Quality
Fast 1024px image · 720p video. Z-Image Turbo for images, Qwen Edit Lightning for edits. Fastest renders.
High Quality 1080p image · 1080p video. Z-Image for images, Qwen Edit 2511 for edits. Sharper detail.
Pro 2048px image · 1080p video. Flux.2 Dev for both images and edits. Top fidelity.

Up at the top-left, the Sogni Creative Agent 1.0 dropdown controls the brain. Five thinking modes, one click each.

Thinking Mode
Auto Default. Decides how long to think on each turn.
Instant Snappy. Answers right away, no reasoning pass.
Thinking Slower, better. Takes a longer route for tougher briefs.
Unrestricted No filters. Drops content restrictions.
Thinking + Unrestricted Both at once. Extended reasoning, no filters.

Tip. Ask for GPT Image 2 or Seedance 2.0 directly in your prompt and Sogni Chat routes to those models even if your quality tier defaults somewhere else.

Default Media Quality dropdown showing Fast (1024px image, 720p video), High Quality (1080p, 1080p), and Pro (2048px, 1080p) options.
Sogni Creative Agent 1.0 model dropdown showing Auto, Instant, Thinking, Unrestricted, and Thinking + Unrestricted modes.
Two dials, top of the chat. Pick once and the rest of the session handles itself.

Write a portrait, or drop in one you already have.

You have two ways into the same workflow. Write the brief, the chat reads natural language, so describe your character the way a director would brief an artist, traits, lighting, mood, colour palette, and Sogni Chat translates that into the right call for the right model.

Or upload a hero you already have. Drag any character image into the composer (or click the + in the chat box) and Sogni Chat will treat it as the source of truth for everything that follows, the storyboard, the panels, and the final Seedance render.

Don't know how to set up a Persona? Take a look at the dedicated walkthrough: Meet your Personas — your face in every render. Once your character is saved, the agent recognises the name and locks the identity across every panel automatically.

That single message is enough. The agent picks the right image model for the quality tier you set, queues the job, and hands you the cost-confirmation panel before a single Spark is spent.

Sogni Chat composer with a long character brief typed in: front-facing bust portrait of a cyberpunk anime heroine with cyan, teal, magenta and violet highlights.
One natural-language brief in the composer. No prompt-engineering required.
Prompt · Character brief
Front-facing bust portrait of an original anime heroine, looking directly at the camera. Long dark high ponytail with neon cyan, teal, magenta and violet highlights. Bright violet eyes, confident expression. Futuristic black crop top and oversized jacket with clean neon accents. Bright soft white-lavender-cyan gradient background, minimal glow, subtle geometric shapes. Clean cyberpunk anime style, vibrant colors, simple design, sharp face details

Every job estimates first. You decide when to spend.

Before any media generation runs, Sogni Chat opens a Confirm generation panel with the exact job estimate, the model (Flux.2 Dev on Pro), the output size, the quality tier, and the cost in Spark credits. For this character bust it's 8.2 Spark, about $0.04 USD.

Want to silence this dialog for the session? Toggle Don't see me again this session. The confirmation budget setting lets you go from Ask every time to fully automatic, in case you trust the agent and want longer chains to run uninterrupted.

Want the bigger picture? Learn more about Sogni pricing, Spark packs, and job estimates before you run a larger workflow.

Confirm generation panel: Flux.2 Dev image at 1248x832, pro quality, estimated 8.2 Spark / $0.04 USD, with Cancel and Start generation buttons.
The job estimate. Every render is metered, transparent, and yours to approve.

The first render lands. From here on, the character stays locked.

A few seconds later, the chat streams in the result, a sharp bust portrait of your cyberpunk anime heroine, jewel-toned hair, violet eyes, the exact jacket and crop top you described. The agent commentates as it goes ("Perfect — your original anime heroine is looking fierce!"), then asks the next question for you: "What are you thinking for her? Want to animate her into a short clip, generate some action poses or different angles?"

Sogni Chat result: the original brief on the right, and on the left a generated bust portrait of a cyberpunk anime heroine with cyan, teal, magenta highlights and violet eyes. The agent suggests animating her or generating more angles.

The character now lives in the conversation's asset manifest. From here on you can refer to "her", "the cyberpunk girl", or "the last image", and the model knows exactly who you mean across every future turn.

Standalone full-resolution view of the cyberpunk anime heroine: bust portrait, dark hair with cyan, teal, magenta highlights, magenta eyes, leather jacket with neon accents.

The streamed result inside the chat, and the full-resolution character on its own. Same hero, ready for the storyboard.

One brief, any panel count, one printable sheet.

Now hand the agent a structured beat list. Describe a story, a routine, or a tonal arc, and tell it how many beats, shots, or panels you want, what aspect ratio you want, and how long the final video should be. Sogni Chat composes a single storyboard sheet, every panel drawn in sequence, with on-screen text overlays for each beat.

Prompt · Example storyboard brief
The main heroine is the same. Create an anime-style storyboard (max duration 15s) showing her self-care morning routine. She has a dark high ponytail with neon cyan, teal, magenta, and violet highlights, violet eyes, and a clean cyberpunk anime look. Show her in multiple panels with a bright soft aesthetic, cozy morning mood, clean composition, soft lighting, and minimal background clutter. The heroine explains what she is doing step by step. Panels: morning mirror, brushing teeth, washing face, foam cleanser, face mask, hair care, skincare serum, moisturizer, making coffee, preparing breakfast, eating breakfast, final fresh look. Keep the character design consistent in every panel. Use expressive but simple poses, clear storytelling, bright bathroom and kitchen interiors, soft pastel tones, subtle neon accents, and a fresh cute self-care atmosphere. Make the dialogue short, natural, and easy to read.

No idea what your storyboard should be? Just ask for a storyboard with this hero and let the chat figure it out, theme, beats, captions, the lot. If you want a specific count or format, say it directly in the prompt, six panels, twenty panels, square, vertical 9:16, widescreen 16:9, or however many beats the story needs. Sogni Chat fills in any missing creative details and hands you back a finished sheet to react to.

Why one sheet instead of separate panel calls? Because rendering all panels together locks the character design across every beat, same face, same hair, same outfit, same lighting language. That consistency is what makes the next step (turning it into video) actually work.

Sogni Chat with a generated self-care morning routine storyboard sheet in the response, captions visible on each panel, and a follow-up prompt asking if the user wants to generate the video.
Close-up of the self-care morning routine storyboard sheet, showing the heroine across multiple beats from morning prep to heading out for the day.
One generation, one shot list. Captions baked in, character locked across every beat, with the panel count set by your prompt.

Hand the sheet to Seedance 2.0. The storyboard becomes the shoot.

Two ways to fire it off. Reply with "Looks good, generate the video.", or just tap the button that appears right under the finished storyboard. Either way the agent feeds the sheet into Seedance 2.0, treats every panel as a keyframe, and stitches motion, transitions, and pacing between them, the same heroine, the same outfit, the same lighting, now actually moving.

Prompt · Render
Looks good, generate the video.

One short line. Seedance does the rest. The chat confirms "Successfully created the Seedance video using the attached media assets." and drops the finished clip straight into the timeline, native audio included.

Want a different feel? Swap one word in the prompt and the agent re-routes to LTX 2.3 or Wan 2.2 instead. Same storyboard, different render rhythm, the character stays locked the whole time.

Sogni Chat with the storyboard sheet, the user message 'Looks good, generate the video.', and the resulting Seedance 2.0 video player embedded in the conversation, followed by the message 'Successfully created the Seedance video using the attached media assets.'
The chat conversation, and the finished Seedance 2.0 render straight from the sheet. Unmute from the player controls for sound.
Optional · Step 10

Hit a snag? Tap the three dots and tell us.

Every chat in chat.sogni.ai ships with an in-app feedback shortcut. If something looked off, a render broke halfway, the agent picked the wrong model, the storyboard skipped a beat, you don't need to leave the conversation, switch tabs, or hunt for a support page. The team can see exactly what you saw.

In the top-right corner, right next to the Default Media Quality badge, you'll find a small three-dot icon. Tap it and a compact menu drops down with four entries: Show Media, Report a bug or give feedback, Export Chat (.zip), and New Chat.

Pick "Report a bug or give feedback" and a clean modal opens. One free-text field, one toggle, two buttons. Tick "Submit this conversation and all its uploaded and generated assets" so the team gets the full picture, hit Submit, done.

Why it matters. Bug reports with the original conversation attached take a fraction of the time to triage, the team sees the exact prompts, the exact assets, and the exact route the agent took. The fastest fixes always start here.

Sogni Chat top-right corner: Default Media Quality Pro badge next to a three-dot menu, opened to show Show Media, Report a bug or give feedback, Export Chat (.zip) and New Chat options.
Report a bug or give feedback modal: textarea labelled 'Describe what happened or what could be better', a checkbox to submit the conversation and assets, Cancel and Submit buttons.
Three dots Menu Modal Submit The team sees the same chat you saw.

One chat. One hero. From description to a storyboard to a finished video.

That is the whole point of Sogni Chat. The same conversation that drafts your character drafts the sheet, the same conversation that drafts the sheet drafts the video. Every step is metered, every step is reversible, and the asset manifest keeps your character locked across all of it.

Start with a sentence, end with a clip. Then save it as a Workflow Template and the next morning routine, product demo, or social spot is one prompt away instead of a pile of separate panel calls.