Workshop

Every model. Every tool. Every tradeoff.

Other tools hide which model rendered your clip. Hybrig shows you the whole stack — which model runs where, what it's good at, what it's mediocre at, and what it costs. Pick the right tool for the job, or trust the recipe library to pick for you.

10 models8 pipeline tools12 recipes

Section 01 · Compare

Compare these models

Pick up to four models. Honest scores, honest costs. The 0-100 marks are our internal calibration — they won't flatter any vendor.

Pick up to 4 models to compare

Tell me what I'm making

	Seedance 2.0 Standard	Seedance 2.0 Fast	Wan 2.1 (local)
Cost / 5s clip	$1.51	$1.21	free
Wall clock / 5s	~2m	~2m	~3m
Face quality (0-100)	92	85	74
Motion quality (0-100)	85	78	65
Identity preservation	90	85	78
Cinematic look	82	72	60
Hardware	cloud	cloud	16GB VRAM
Free?	no	no	yes
Lip-sync	yes	yes	no
Recipes	Premium Brand Avatar, Multi-Take A/B, Brand Ad (Hermes-style)	UGC Spokesperson (Polished), Multi-Take A/B, Quick Social Clip	UGC Spokesperson (Free Local), UGC Spokesperson (Polished), Quick Social Clip, Animated Explainer

Seedance 2.0 Standard

Cost / 5s clip: $1.51
Wall clock / 5s: ~2m
Face quality (0-100): 92
Motion quality (0-100): 85
Identity preservation: 90
Cinematic look: 82
Hardware: cloud
Free?: no
Lip-sync: yes
Recipes: Premium Brand Avatar, Multi-Take A/B, Brand Ad (Hermes-style)

Seedance 2.0 Fast

Cost / 5s clip: $1.21
Wall clock / 5s: ~2m
Face quality (0-100): 85
Motion quality (0-100): 78
Identity preservation: 85
Cinematic look: 72
Hardware: cloud
Free?: no
Lip-sync: yes
Recipes: UGC Spokesperson (Polished), Multi-Take A/B, Quick Social Clip

Wan 2.1 (local)

Cost / 5s clip: free
Wall clock / 5s: ~3m
Face quality (0-100): 74
Motion quality (0-100): 65
Identity preservation: 78
Cinematic look: 60
Hardware: 16GB VRAM
Free?: yes
Lip-sync: no
Recipes: UGC Spokesperson (Free Local), UGC Spokesperson (Polished), Quick Social Clip, Animated Explainer

Section 02 · Model Atlas

Every video model we support

Each card is a real review. We're especially direct about weaknesses — knowing what a model is bad at is more useful than knowing what it's good at.

Showing 10 of 10

Seedance 2.0 Standard

standard · cloud

$0.30/sec

sample seedance-2.0-standard

Top-tier quality. Best face lock + VFX. Use for finals.

Strengths

+Best face lock under heavy motion in our tests
+Strong cinematic VFX and lighting consistency
+Reliable identity preservation across takes

Weaknesses

−Mediocre on hands when subject gestures
−Backgrounds can drift on clips longer than ~6s

5s clip

$1.51

~Wall clock

Hardware

cloud

Best for

Premium Brand AvatarMulti-Take A/BBrand Ad (Hermes-style)

Seedance 2.0 Fast

fast · cloud

$0.24/sec

sample seedance-2.0-fast

~20% cheaper, slight quality dip on complex scenes. Drafts + iteration.

Strengths

+~20% cheaper than Standard with most quality intact
+Faster turnaround — good for A/B drafting
+Same face-lock backbone as Standard

Weaknesses

−Visible quality dip on complex multi-subject scenes
−Texture detail softer than Standard at 1080p

5s clip

$1.21

~Wall clock

Hardware

cloud

Best for

UGC Spokesperson (Polished)Multi-Take A/BQuick Social Clip

Wan 2.2 (local)

standard · local

LOCAL FREE

sample wan-2.2-local

Alibaba's Wan 2.2 — the new default local model. Slightly better motion + identity than 2.1, same install path. Runs on your own GPU via ComfyUI + WanVideoWrapper. ~3-8 min per 5s clip on a 4090. Free forever. Requires ~14 GB on disk and 16+ GB GPU memory.

Strengths

+Free forever — runs on your own GPU, no per-second cost
+No cloud queue — starts in seconds, not hours
+Privacy: footage never leaves your machine
+Better motion + identity than Wan 2.1 in our tests

Weaknesses

−Still behind Seedance on demanding face-heavy shots
−Hands and crowd scenes weaker than top cloud models
−Requires ComfyUI install and 16+ GB GPU memory

5s clip

free

~Wall clock

Hardware

16GB VRAM

Best for

UGC Spokesperson (Free Local)UGC Spokesperson (Polished)Quick Social ClipAnimated Explainer

Wan 2.1 (local)

standard · local

LOCAL FREE

sample wan-2.1-local

Older Wan generation — kept around as a backup for when 2.2 misbehaves on your install. Same look, slightly weaker motion + identity than 2.2. Runs on your own GPU via ComfyUI + WanVideoWrapper. ~3-8 min per 5s clip on a 4090. Free forever.

Strengths

+Free forever — runs on your own GPU, no per-second cost
+No cloud queue — starts in seconds, not hours
+Privacy: footage never leaves your machine
+Mature: well-tested workflow, lots of community fixes

Weaknesses

−~70-80% quality vs Seedance on demanding shots
−Hands and complex motion are noticeably weaker
−Requires ComfyUI install and 16+ GB GPU memory

5s clip

free

~Wall clock

Hardware

16GB VRAM

Best for

UGC Spokesperson (Free Local)UGC Spokesperson (Polished)Quick Social ClipAnimated Explainer

HunyuanVideo (local)

standard · local

LOCAL FREE

sample hunyuan-video-local

Tencent's HunyuanVideo — strong on natural motion, especially body movement and physics. Runs locally via ComfyUI's native Hunyuan nodes. Use the smaller, slightly less accurate version (fp8) — about 13 GB on disk. Needs 16+ GB GPU memory. Slower than Wan: plan on 6-12 min per 5s clip on a 4090.

Strengths

+Best-in-class motion realism among free local models
+Body physics and limb movement noticeably better than Wan
+Strong on action, walking, gestures

Weaknesses

−Slower than Wan — heavier model
−Identity lock weaker than Wan on tight close-ups
−Larger download (~13 GB)

5s clip

free

~Wall clock

Hardware

16GB VRAM

Best for

Action DemoUGC Spokesperson (Polished)Quick Social Clip

LTX-Video (local)

fast · local

LOCAL FREE

sample ltx-video-local

Lightricks LTX-Video — small, fast, lower quality. About 9 GB on disk, runs on as little as 12 GB GPU memory. Renders a 5s clip in roughly a minute on a 4090. Use it for drafts, blocking ideas, or when bigger models choke on your install.

Strengths

+Very fast — ~1 min per 5s clip on a 4090
+Small download (~9 GB), runs on 12 GB GPU memory
+Great for iterating on direction before committing to a slow model

Weaknesses

−Visibly lower quality than Wan 2.x or Hunyuan
−Identity preservation weaker on close-ups
−Less detailed motion — best for simple shots

5s clip

free

~Wall clock

Hardware

12GB VRAM

Best for

Quick Social ClipAnimated Explainer

Kling 2.1 (cloud)

pro · cloud· coming soon

$0.50/sec

sample kling-2.1-cloud

Kuaishou's cinematic model — different aesthetic, sometimes wins on cinematic shots.

Strengths

+Distinctive cinematic look — strong on establishing shots
+Good camera-move physics (push-ins, dolly drifts)
+Lighting feels more 'film' than other cloud models

Weaknesses

−Identity preservation drifts on tight close-ups
−Pricier than Seedance Standard with worse face lock

5s clip

$2.50

~Wall clock

Hardware

cloud

Best for

Cinematic Cold OpenBrand Ad (Hermes-style)

Hailuo 2.0 (cloud)

pro · cloud· coming soon

$0.48/sec

sample hailuo-2.0-cloud

MiniMax's video model. Strong on motion realism.

Strengths

+Best-in-class motion realism for action / sports
+Handles fast camera moves without warping
+Crowd / multi-subject scenes hold up

Weaknesses

−Face lock weaker than Seedance on close-ups
−Inconsistent lighting frame-to-frame on long clips

5s clip

$2.40

~Wall clock

Hardware

cloud

Best for

Action DemoCinematic Cold Open

Wan 2.2 Animate (cloud)

pro · cloud· coming soon

$0.40/sec

sample wan-2.2-animate-cloud

Alibaba's Wan 2.2 Animate — image+motion-to-video, good for reference-driven animation.

Strengths

+Reference-driven animation — strong from a single image
+Storyboard-friendly: respects pose / motion guides
+Good for explainer / illustrative work

Weaknesses

−Less photoreal than Seedance on talking heads
−Pricing and slug not yet verified — treat as experimental

5s clip

$2.00

~Wall clock

Hardware

cloud

Best for

Animated ExplainerCustom Workflow Builder

Google VEO 3.1 (cloud)

pro · cloud· coming soon

$0.50/sec

sample veo-3.1-cloud

Google's VEO 3.1 — primarily text-to-video with optional image reference. Strong cinematic output.

Strengths

+Best-in-class text-to-video coherence
+Cinematic look with very good world physics
+Native synced audio on supported endpoints

Weaknesses

−Reference-image fidelity weaker than Seedance
−Identity-lock not as tight on returning characters

5s clip

$2.50

~Wall clock

Hardware

cloud

Best for

Cinematic Cold OpenAnimated Explainer

Section 03 · Tool Atlas

Everything else in the pipeline

A finished render isn't just a model call. Voice cloning, identity verification, polish, captioning, brand frames — all the tools that participate, what they cost, where they run.

Pre-render

pre-render · voice

ElevenLabs

ElevenLabs Inc.

live

Voice cloning, TTS in 30+ languages, and lip-sync polish. We use it as the spokesperson's voice and as the 'translator' for multilingual dubs.

Runs on: Cloud API
Cost: ~$0.30 per 1k characters (Creator tier). Cloning included in monthly plan.

pre-render · identity

Wan ClipVision

live

Face encoder that locks the spokesperson's identity into the diffusion model. Runs as part of the local Wan workflow — no cloud round trip.

Runs on: On your GPU
Cost: Free. Bundled with WanVideoWrapper.

Post-render

post-render · polish

Topaz Astra

Topaz Labs

planned

AI upscale + temporal artifact cleanup. Takes a 720p Wan render to a clean 1080p/4K. Fixes flicker on long clips and softens compression artifacts.

Runs on: Desktop app
Cost: $299 one-time desktop license. No per-render fee.

post-render · polish

Hedra Character-3

Hedra Labs

experimental

Audio-driven micro-expressions on the talking head — eyebrow raises, head tilts, breath. Layered after the base render to add humanity.

Runs on: Cloud API
Cost: Roughly $0.15/sec when integrated.

post-render · verification

Identity Verifier

live

Hybrig's own face-match scoring across takes. Flags any take where the rendered face drifts more than ~25% from the source character pack and offers an automatic re-roll.

Runs on: On your GPU
Cost: Free. Runs on CPU in the worker.

Export

export · editing

DaVinci Resolve

Blackmagic Design

live

Final color grade, audio sweetening, and timeline assembly. Hybrig exports cut-ready clips that drop straight into a Resolve project.

Runs on: Desktop app
Cost: Free (Resolve free version covers everything we need).

export · compositing

FFmpeg / CapCut

live

Caption burn-in, aspect ratio reframing (9:16 / 1:1 / 16:9), and platform-specific export presets. Runs server-side via fal.ai or locally via the worker.

Runs on: Hybrig server
Cost: Free. Server-side ffmpeg costs us only fal compute cents per export.

export · compositing

HyperFrames

live

Branded intro/outro/chyron compositor. Lets you stamp consistent brand frames onto every render with a single template.

Runs on: On your GPU
Cost: Free. Built into Hybrig.

Time

varies

Use this recipe

Section 05 · Decision Tree

Not sure where to start?

Two questions, one recommendation. Skip the atlas if you just want to ship something today.

Step 1 — What are you making?