36:14
Higgsfield
The video is a full beginner-oriented walkthrough of Higgsfield Cinema Studio's upgraded interface, mapping the entire pipeline — project setup, image generation (Soul Cinema vs. Cinema 2.0), the three video-generation modes (single-shot, multi-shot auto, multi-shot manual), character/location/prop creation via the director's panel, and voice cloning — for users who left the tool and returned to find new buttons.
The interface is organized into a project → image → video → audio pipeline, navigated from a persistent top (and mirrored bottom) header.
Image generation happens first and locks in aspect ratio, 'film look,' and any character/location/prop references used downstream in video generation.
Video generation offers three modes: single-shot animation, multi-shot auto (AI selects shots), and multi-shot manual (user controls each shot's duration on a timeline).
Genre selection is functional, not cosmetic: it drives the AI's automatic choices on shot selection, camera angles, pacing, and speed-ramp behavior.
Two image models exist with different tradeoffs: Soul Cinema (preset-driven, simple prompts, capped at 2K) and Cinema 2.0 (more manual control, grid generation, up to 4K).
A 'color signature' feature extracts a palette from an uploaded reference image and applies it to every image generated afterward in the session.
Characters, locations, and props are built once via the director's panel ('Build your cast', 'save location', 'save as element') and reused through an @-tag referencing system.
Prompts are capped at 3 tags total, and characters/locations/props all compete for that same 3-tag budget.
Start frame is required for single-shot generations that reference characters, but not for multi-shot generation, which instead risks environment/outfit inconsistency without one.
Clip duration (5–12s) scales credit cost, and multi-shot manual is hard-capped at 6 scene generations per video regardless of total duration.
An audio (voice) feature can lock a character to one consistent AI voice across every clip, with automatic lip-sync, and supports cloning the user's own voice from a recording.
The creator explicitly prefers multi-shot manual over multi-shot auto, describing the auto mode's output as unreliable.
Project Panel — The left-hand list of all saved Higgsfield projects, from which a new project starts with a blank slate via the plus button while existing projects remain intact and can be edited, moved, or pinned via right-click. Apply: Use the plus button to start a fresh project without losing prior work, and right-click any project to rename, reorder, or pin it for quick access.
Top Header Navigation — A persistent top (and mirrored bottom) navigation bar that moves the user between the Image, Video, and Audio stages of a project. Apply: Move sequentially from Image to Video to Audio rather than hunting for buried menu options.
Image Tab / Film Look — The first-step workspace where the initial frame of a shot and its overall 'film look' are generated before any animation happens. Apply: Generate and lock in the first frame's look here before moving to the video tab to animate it as a keyframe.
Video Tab: Single-Shot Animation — A video generation mode that animates one continuous shot from a chosen start (and optionally end) frame. Apply: Use it when a scene needs one uncut, controlled camera move rather than several intercut angles.
Video Tab: Multi-Shot Auto — A video generation mode where the AI automatically selects shots, cuts, and pacing to fill the chosen duration based on genre. Apply: Reserve for quick drafts, since the creator reports it 'gives me stuff that I don't need' and prefers manual control instead.
Video Tab: Multi-Shot Manual — A video generation mode where the user drags a timeline to set each shot's individual duration and supplies a separate prompt per shot, capped at 6 generations per video. Apply: Build multi-cut scenes by allocating duration per shot on the timeline and writing a distinct prompt for each of up to 6 shots.
Aspect Ratio Lock at Image Stage — The output aspect ratio is set when generating the image, not the video, and video generation inherits it automatically. Apply: Pick the target aspect ratio (e.g., 16:9 or 21:9) during image generation so every downstream video shot frames correctly.
Quality Settings (1080p/720p) — A video-output resolution toggle offered at the video generation stage. Apply: Choose 1080p or 720p output depending on the intended final delivery quality.
Genre-Driven Auto Generation — A genre selector (action, horror, comedy, detective, fantasy, suspense, etc.) that informs the AI's automatic choices for shot selection, camera angles, pacing, and speed ramping. Apply: Set the genre before using multi-shot auto or speed-ramp 'auto' so the AI's stylistic choices match the intended tone (e.g., action genre triggers fight-scene speed ramps).
Audio On/Off Toggle — A per-clip switch that disables generated audio, which also lowers the credit cost from 20 to 15. Apply: Turn audio off for b-roll shots meant to sit under external music or dialogue to save credits.
Batch Generation — A parameter that generates multiple different output variations from the same prompt in a single generation cycle. Apply: Set batch count to 2 or more to get several option variants from one prompt without re-queuing generations one at a time.
Start Frame — A reference image (generated or a saved character/location/prop preset) used to animate a shot from a fixed beginning point; required for single-shot generations using character references, and applies globally across a whole multi-shot video rather than per-shot. Apply: Generate or select the exact opening image first, then feed it in as the start frame so the animation begins from that precise look.
End Frame — A specified target image that the AI generates a continuous shot toward, ending on that frame. Apply: Supply an end frame when a shot needs to land on a specific pose or composition by its final moment.
Camera Movement Presets — A library of movement styles (handheld, static, dolly shot, dolly left, etc.) with hover-to-preview, applied to a generated shot. Apply: Hover over each movement option to preview it, then select the one matching the intended camera language for the shot.
Speed Ramp Modes — Pacing controls — impact (ramps speed up to 150–250%), slow motion, linear, custom, and auto — that alter a shot's playback speed over its duration, with 'auto' influenced by the chosen genre. Apply: Apply 'impact' for a sudden speed surge in action beats, or 'slow motion' to decelerate a moment, choosing 'auto' only when the genre setting matches the desired effect.
Clip Duration & Credit Scaling — Clip length is adjustable from 5 to 12 seconds, with credit cost scaling accordingly (6 credits for 3 seconds up to 24 credits for 12 seconds). Apply: Balance shot length against credit budget, using shorter clips for cheaper iteration and reserving longer clips for shots that need it.
Soul Cinema (image model) — The newest image generation model in Higgsfield, which applies cinematic presets automatically from simple prompts and caps out at 2K resolution. Apply: Use Soul Cinema for fast, cinematic-looking stills from minimal prompting when 2K resolution is sufficient.
Cinema 2.0 (image model) — An alternate image generation model offering grid-style output and more manual control, capable of generating up to 4K resolution. Apply: Switch to Cinema 2.0 when higher resolution (up to 4K) or a grid of options (2×2/3×3) to compare is needed.
Color Signature — A feature that extracts a color palette from an uploaded reference image (e.g., a still from The Mask, 1994) and applies it to all subsequently generated images in the session. Apply: Upload a reference image whose look you want to match, apply its color signature, and it will persist across every image generated afterward in that session.
Image Editing Suite — A set of post-generation image tools including upscale, enhance, relight (adjust brightness/darkness), and angle-swap. Apply: Run a generated image through upscale/enhance for quality, relight to fix exposure, or angle-swap to get a different camera angle of the same image.
Camera + Lens Preset System — A set of selectable camera/lens presets (e.g., 50mm at F1, vintage prime lens) that alter optical characteristics such as depth of field and flare. Apply: Pick a lens preset like the vintage prime lens to add flare effects, or set focal length/aperture (e.g., 50mm F1) for a specific depth-of-field look, then save the combination as a reusable preset.
Color Grading — A per-image toggleable control set (hue, saturation, contrast) with shortcut presets like 'soft skin' and 'old lens.'. Apply: Toggle color grading on and either fine-tune hue/saturation/contrast manually or apply a preset like 'soft skin' for a quick stylistic pass.
@ Tag Referencing System — Typing '@' opens a menu of previously saved characters, locations, and props (or lets you create a new one) to insert into a prompt, capped at 3 tags per prompt. Apply: Write the scene description first, then use '@' to tag in up to 3 saved elements (mixing characters/locations/props) rather than describing them from scratch each time.
Grid Generation (2×2 / 3×3) — An option in Cinema 2.0 to generate a grid of 4 (2×2) or 9 (3×3) image variations at once for side-by-side comparison. Apply: Use a 2×2 or 3×3 grid when you want several visual options to choose from in a single generation pass.
Location Tab / Save Location — A simplified location-creation workflow (without full color-palette control) whose outputs can be saved and renamed into the director's studio folder for reuse. Apply: Generate and save a location once, then reuse it across multiple scenes by name instead of regenerating the environment each time.
Build Your Cast (Character Creation) — A parameterized character-creation workflow inside the director's panel that sets genre, budget ($10M–$500M), era (e.g., 1920s), archetype (hero/explorer/lover/creator), identity, physical appearance, and outfit; genre choice also tints the character's overall appearance. Apply: Build a reusable cast member by stepping through genre, budget, era, archetype, identity, appearance, and outfit rather than describing the character freshly in every prompt.
Save as Element — A save action that stores a generated character (or other asset) into the project's element database for later @-tag reuse. Apply: Save a satisfactory character generation as an element immediately so it can be tagged into future prompts instead of regenerated.
Director's Panel — A back-end management view listing every parameter used to generate a character/location/prop, including the ability to rename an asset (default auto-assigned names can be changed only from here). Apply: Open the director's panel to rename an auto-named character or to review/adjust the exact parameters that produced a saved asset.
Prompt-First, Then-Tag Workflow — The creator's recommended prompting order: write the scene's descriptive prompt text first, then attach @-tagged character/location/prop references, since skipping the tags leaves the prompt uninformed about which specific elements to use. Apply: Draft the full scene prompt first, then tag in the specific saved characters/locations/props needed, rather than tagging before writing the description.
Voice / Audio Feature (Character Voice Consistency) — An audio system that, by default, assigns a different AI voice to a character on every generation, but can be locked to one consistent voice (with automatic lip-sync) selected once and reused across all of that character's clips. Apply: Pick and lock a single voice for a character (e.g., 'Sloan') so every clip featuring them uses the same voice instead of a new random one each time — most worth doing for intimate, dialogue-heavy back-and-forth scenes rather than action/minimal-speech content.
Voice Cloning Workflow — A process for giving a character the user's own (or any custom) voice by recording a sample live or uploading an existing voice note/podcast recording, cloning it, and saving it to the voice preset list. Apply: Go to voice presets → create custom voice → record or upload a voice sample → save the clone, then select it from the preset list whenever that character speaks.
The audio toggle is not just a mute switch — disabling it actively lowers the credit cost from 20 to 15, so the creator recommends turning it off for b-roll meant to sit under external music or dialogue.
The 3-tag cap, not prompt length, is framed as the real ceiling on scene complexity, since a single prompt can reference at most 3 saved characters/locations/props combined.
The creator's recommended workflow order (write the prompt text first, then tag elements with @) implies tags are appended references rather than parsed at their position in the sentence.
Still-image resolution and video-output resolution are decoupled and capped differently (stills up to 4K, video capped at 1080p), which is why the creator settles on 2K as a deliberate 'balance' rather than always maximizing image quality.
Multi-shot manual's hard ceiling of 6 generations per video means longer runtimes are achieved by lengthening individual shots, not by adding more cuts.
Voice consistency is framed as mattering mainly for intimate, dialogue-heavy back-and-forth scenes; in action-heavy or ambient-audio content the creator says viewers usually can't tell a character's voice changed between clips anyway.
The 'animate' shortcut reachable directly from image generation is flagged as a trap to avoid, since it exits Cinema Studio entirely instead of staying in the guided pipeline.
«So, don't worry, stick around and let's go through it together.»
— 00:11
«So, if you do two, it's going to be the same prompt and it's going to give you two different options at once, so you just have to wait one time instead of generating it one at a time.»
— 03:02
«Soul Cinema is the newest addition to the Hicksfield ecosystem. It's going to by default give you all the presets you need to have a cinematic image right from the beginning with very simple prompts.»
— 07:46
«There's a lot of controls and a lot of things that we can get into. And if you don't want to mess with it individually, they have presets.»
— 12:32
«if you just do those three things, you're not informing the prompt what to do»
— 20:04
«I haven't had luck with the auto multi-shot. It gives me stuff that I don't need»
— 25:40
«you can only do three tags»
— 28:21
«or you can upload yourself from a voice note or podcast that you did, and that will take your voice and clone it, and that's how you can have a character with your own voice.»
— 35:27
«So, if you want to keep up with everything that goes on in the world of Hexfield and AI filmmaking, drop a subscribe, drop a like, drop a comment, and I'll see you guys next time.»
— 35:56
Reception
Viewers appreciated the comprehensive guidance and found it genuinely helpful, but concerns about incomplete feature coverage, platform UI changes, and unclear pricing guidance tempered enthusiasm.
This is a comprehensive product-tutorial map of Higgsfield Cinema Studio's workflow rather than a critical evaluation of it, treating each feature as a given and occasionally volunteering the creator's own usage preferences (e.g., manual over auto multi-shot) as practical tips rather than as criticism of the tool.
36:14