Tripo Retopology v3 rebuilds dense or AI-generated meshes as clean low-poly topology, with triangle or quad output, a 500-face floor and optional Smart UV unwrap.
Models
All Models
Build cinematic scenes from up to 15 images and 3 voice references. Generate 3 to 16 second videos with synchronized audio, five aspect ratios and resolutions up to 4K.
Bring a single image to life with cinematic motion, transformations and synchronized audio. Create 3 to 16 second videos from 540p to 4K for products, characters and worlds.
Lighter Kandinsky 6.0 super-resolution for short videos: 2x, 2.25x or 4x upscale, 480p to 1080p at 2.25x, photoreal or animated, original audio kept. Clips up to 5 seconds.
Upscale short videos 2x, 2.25x or 4x with Kandinsky 6.0 super-resolution: 480p to 1080p at 2.25x, sharper detail on photoreal and animated clips, audio kept, up to 5 seconds.
Bring a still to life as a 5-second 480p clip with native synced audio: lip-synced dialogue, foley, ambience and music. Kandinsky Lab's 29B Pro tier with guidance control.
Cinematic 5-second 480p video with native synced audio from text: lip-synced dialogue in English or Russian, foley, ambience and music. Kandinsky Lab's 29B flagship tier.
Animate a still into a 5-second 480p clip with native synced audio: lip-synced speech, foley, ambience and music. Kandinsky Lab's fast 3B tier, keeps identity from frame one.
Text to 5-second 480p video with native synced audio: lip-synced dialogue, ambience, foley and music in one pass. Kandinsky Lab's fast 3B tier for drafts, social and volume.
Edit or combine up to 10 reference images in one prompt, tagged @Image1 to @Image10, into up to 9 results at 1K, 2K or 4K, with auto aspect ratio and character elements.
Kling's Omni image model: up to 9 images per run from text at 1K, 2K or native 4K, in 8 aspect ratios, with face-controlled elements for consistent characters.
Transform a reference image with a text prompt: swap materials, relight, restyle or restage it, up to 9 results per run at 1K or 2K, with optional face-controlled elements.
Generate up to 9 images per run from text at 1K or 2K across 8 aspect ratios, with face-controlled elements that keep up to 3 characters consistent in any new scene.
Continue any video from its last frames: add 1 to 15 seconds of new action at 360p to 1080p, with optional generated audio and anime, clay, comic or cyberpunk styles.
Edit a video with a text prompt: relight, restyle, swap subjects or environments while the original motion, timing, framing and audio stay intact. Up to 15s, 720p.
Insert a new scene into a video with MiniMax H3 Max: guided by a prompt, reference images or videos, then return to the original footage. 5 to 13 seconds, native audio.
Recast a video with MiniMax H3 Max: swap up to 4 people for new ones from photos, keeping the original motion, camera, cuts and audio, at 768P or 1080P.
Extend any video with MiniMax H3 Max: describe what happens next and add 1 to 15 seconds of new footage with native audio, from 480P up to 2K, as one clip.
Extend any video faster with MiniMax H3 Max Turbo: describe what happens next and add 1 to 15 seconds of new footage with native audio, from 480P up to 2K.
Create native 4K images with sharp text and precise layouts. FLUX.3 Image also edits and combines up to 10 references for campaigns, product visuals, and game art.
Low-latency ElevenLabs v4 for fast iterations: the same audio tags and 100+ languages, generated 2 to 4 times faster. Ideal for drafts, voice agents and rapid script tests.
Turn text into expressive speech in 100+ languages, directed with inline audio tags like [whispering] or [laughing]. ElevenLabs' highest-quality voice model, 21 voices.
Edit an image in place and leave the rest untouched: swap text, recolor, add or remove objects, with an optional mask and up to four reference images (three with a mask).
Fast, affordable Seedream 5.0: generate or edit 1K to 2K images with sharp in-image text, up to 10 reference images, and one flat price at every resolution.
Split one finished image into a clean base plus up to 16 transparent PNG layers, rebuilding the background behind each element. Auto, text, or bbox targeting.
Fast, lightweight Gemini text-to-speech for high-volume voiceover, voice agents and read-aloud: 30 voices, 24 languages, inline vocal tags and two-speaker dialogue.
Turn any song into a full AI music video. Choose from 15 visual styles or a custom reference, add a character photo and lyrics, export up to 1080p in five aspect ratios.
Generate a fully textured 3D model from a text prompt alone. New geometry resolution tiers (standard, 2K, 4K) add finer surface detail than Meshy 7, with 2K-8K PBR texturing.
Turns a still portrait into a natural talking video synced to your own audio track. Feed an image and 5-15s of speech, get realistic mouth movement up to 2K resolution.
Recraft's fastest 1K raster model. Strong prompt accuracy and integrated text rendering at a lower cost than V4.1. No style controls: rejects style IDs and reference images.
Build 3D objects from text or up to 8 images, or edit an existing model with visual guidance. Choose effort, topology and export format, with a Blender file included.
Detects scene cuts in raw footage and extracts up to 200 quality-filtered clips automatically, dropping shaky, static, blurry, or duplicate shots for you.
Turns an image into a clean, editable SVG. Made for icons, logos, stickers, flat illustrations and line art: crisp edges, exact colors, smooth curves.
Extend or rework an existing track: keep the sections you want unchanged and generate new verses, intros, or outros around them, in ElevenLabs' own voice and style.
Compose full songs section by section: set styles, lyrics, duration, and context adherence per part, then let ElevenLabs assemble a cohesive track.
Generate full music tracks from one text prompt: any genre, vocal or instrumental, 3 to 600 seconds, with ElevenLabs' most accurate prompt adherence yet.
P-Video 2 Pro turns a text prompt or first/last frames into video with native dialogue, music and SFX, up to 15s at 480p or 768p, in Cost, Speed or Quality mode.
Birefnet Background Removal runs the open-source BiRefNet model for clean transparent cutouts, holding on to fine edges on hair, fur, and thin straps.
Turn text prompts into cinematic 5-15s videos at 480p or 768p, faster than ever. Fal's MiniMax H3 Max Turbo delivers stronger prompt adherence for quick, on-model video.
Animate a still image into a cinematic 5-15s video at 480p or 768p, faster than ever. Fal's MiniMax H3 Max Turbo adds stronger prompt adherence, plus end-frame control.
P-Video Edit by Pruna AI edits an existing video from a text prompt, optionally guided by up to four reference images for identity or style. Draft mode for fast, low-cost previews.
Convert any SDR video to true HDR: HDR10, HLG, or HDR ProRes up to 4444, plus 16-bit EXR sequences. Works on output from any model, not only Runway, up to 40 seconds.
Tripo P2.0 text to 3D. Native 3D diffusion with clean topology, optional quad mesh output, negative prompt, and face counts up to 50,000 (25,000 with quad).
Tripo P2.0 Multiview (front + left/back/right) to 3D model. Production-grade native 3D diffusion with clean topology, optional quad mesh, and face counts up to 50,000.
Subject-consistent reference-to-video generation from images, motion clips, and audio cues. Fal's tuned MiniMax H3 Max, up to 15 seconds at 768P with native stereo audio.
Edit images from up to 10 reference photos with this instruction-following edit model. Edits: restoration, style transfer, background swaps, and multi-image blending.
Generate images from prompts with Meta's first image model. Faithful instruction-following, and strong multi-step prompt, for posters, UI, and infographics.
Turn a single photo of a place into a navigable 3D Gaussian-splat world. Works indoors or outdoors. Delivers a compact .spz preview plus a downloadable .ply.
Generate editable SVG vector images that follow a reusable Recraft style ID or attached reference images. Built for logos, icons, mascots, and other vector assets.