The melting image is one of the most recognizable visual gestures in surreal art — from Dalí's softening clocks to dripping horror film posters — and it communicates disorientation, transformation, and psychological weight in a single sequence. The AI melt effect on Polyfaced takes an image you upload and uses Kling 2.1's image-to-video model to animate it into a genuine melting sequence: surfaces soften, edges lose their rigidity, and material flows downward following gravity and material-consistent physics. A portrait face melts from the hairline down. A product photo warps at the corners first before the center collapses. A landscape dissolves from the horizon inward. The output is a downloadable MP4 that slots directly into horror edit timelines, psychedelic music video sequences, surreal concept art trailers, and any format where physical impossibility is the point. The melting animation is generated from the actual image you provide — not a generic liquify filter applied uniformly — which means the texture, color behavior, and flow direction reflect the specific visual content of your input. 9:16 vertical output targets Reels and TikTok natively. 16:9 handles YouTube and presentation formats. Image-to-video generation costs 12 credits per 5-second clip, with failed jobs returned to your balance automatically.
What the AI melt effect does
Material-aware melt physics derived from image content
A generic liquify filter applies the same warp across the entire frame regardless of what the image depicts. Kling 2.1 generates the melt sequence by modeling the material properties implied by the visual content — skin and soft tissue flows differently from metal, wax has distinct drip formation characteristics, paint separates in layers, glass sags before it runs. The model infers these behaviors from the trained relationship between visual texture and physical deformation, which is why a melting portrait face looks different from a melting product image even when using identical prompt language. The directionality, viscosity approximation, and rate of collapse differ per material type. This material-sensitive behavior is what separates the output from a filter and makes it readable as that specific thing melting rather than a generic distortion applied to any source.
Gravity and flow direction follow scene geometry
Melt animations that do not account for the orientation of the scene look immediately artificial — material flowing sideways, pooling at the wrong edge, or defying the implied vertical axis of the image. Kling 2.1 generates the flow direction relative to the scene's implied spatial geometry: melt moves downward relative to the horizon line the model reads from the image content, which means an upright portrait produces top-to-bottom flow while a tilted composition adjusts the flow vector accordingly. Drips form at protrusions — a chin, a shelf edge, a headline letterform's crossbar — before running down the surface. Material pools at the bottom of the frame at a rate consistent with the viscosity the model infers from the source texture. The result is a melt sequence that feels physically grounded even while depicting an impossible event.
Color saturation and bloom behavior during melt progression
As solid forms lose structural integrity in the generated melt sequence, color behavior follows: pigments concentrate where material pools, creating deeper saturation at drip tips and accumulation points, while the source regions that have melted away show the lighter, drained hue of the base material. On images with strong color contrast — a red jacket against a white background, a neon sign on a dark wall — the melt sequence generates chromatic bloom at the interface between the melting region and the intact surface. This bloom behavior adds visual richness to the transition and contributes to the psychedelic quality of the output without requiring a separate post-processing step. The color dynamics emerge from the generation rather than being composited on top of it.
Scale-independent: works on faces, products, text, and environments
The AI melt effect is not constrained to a single content category. Portrait photos produce face-melt sequences where the facial structure softens and runs. Product images generate object-melt outputs where the subject loses form in a way that reads as material-appropriate — wax, metal, paint, or plastic each behave distinctly. Typography and graphic compositions melt in ways that preserve readability through the first half of the sequence before full dissolution. Environmental shots — architecture, landscapes, interiors — generate wide-frame melt sequences where the scene dissolves from selective focal points outward. The common thread across all input types is that the melt behavior derives from what the image actually depicts, producing outputs that feel specific to the source rather than generic. This range makes the effect usable across horror, art direction, product advertising, and experimental video formats from a single generation workflow.
How to create a melting animation
- 1
Sign in and open the video studio
Go to polyfaced.com and sign in with Google. New accounts receive 5 free credits automatically — enough to run one test 5-second image-to-video generation and evaluate the melt effect output before committing to a plan. The AI melt effect uses Kling 2.1 image-to-video, accessible from the video studio.
- 2
Upload the image you want to melt
Upload the source image you want to animate into a melt sequence. Any still image works as input: portrait photos, product shots, graphic designs, illustrations, or photography of objects and environments. Images with strong edges, clear subject-background separation, or high contrast between regions tend to produce more readable melt sequences because the model has distinct boundaries to work with during dissolution. Very low-contrast or heavily blurred input images can reduce definition in the melt flow.
- 3
Write a prompt describing the melt style
Describe the character of the melt you want — material type, speed, directionality, and aesthetic. A functional baseline: 'slow melting, dripping from top to bottom, viscous flow, surreal.' For horror output, push toward faster dissolution with more dramatic pooling: 'rapid melt, face dissolving downward, drips forming at chin and cheeks, dark and unsettling.' For psychedelic music video use, describe chromatic richness: 'melting in vivid saturated colors, slow bloom at melt edges, smooth liquid flow.' Specifying viscosity ('thick' vs 'watery') and speed ('slow' vs 'rapid') has a measurable effect on the generated output.
- 4
Set format and duration
9:16 vertical for TikTok, Reels, and Shorts. 16:9 landscape for YouTube, presentation, or cinematic editing. 1:1 for square social posts or graphic design applications. The 5-second duration covers the onset through mid-melt point for most source images — enough to establish the effect without completing full dissolution. At standard quality (720p), image-to-video costs 6 credits per generation. At 1080p, the cost is 12 credits with a 3–4 minute generation time suitable for final output.
- 5
Download and integrate the melt clip
The generated melt clip appears in the studio panel with a direct download link. The MP4 contains the video track without audio and is compatible with DaVinci Resolve, Adobe Premiere, CapCut, and any editor that accepts H.264. To extend the melt sequence, loop the clip or layer it against a static version of the source image with a time-offset blend. Free and Credit Pack accounts have a 14-day access window for downloads. Pro accounts retain the file on R2 storage for 90 days with a shareable link.
Who uses the AI melt effect
Horror and psychological thriller visual content
Melting imagery carries immediate dread and disorientation — associations with deterioration, loss of form, and the physically impossible that horror content exploits directly. An AI melt effect applied to a close-up portrait generates a face dissolution sequence that functions as a horror transition, a dream sequence visual, or a character-transformation moment without practical effects work. Applied to environmental photography — a house, a room, a face in a crowd — the melt sequence suggests reality breaking down. The clip integrates into horror short film timelines and Reels-format horror content as a standalone visual statement or as a transition between narrative moments.
Psychedelic and surreal music video sequences
Music video production in electronic, experimental, and psychedelic genres uses melting imagery as a visual metaphor for altered states, emotional intensity, or sonic weight. An AI melt effect applied to a performer portrait or to graphic frame elements generates a surreal clip that reads as intentional art direction rather than a filter. The color bloom behavior during melt progression — where pigments concentrate and saturate at drip points — aligns with the high-contrast visual language of psychedelic video work. Generating the melt from artist-specific imagery ties the effect to the visual identity of the release rather than producing generic surreal content.
Product and brand campaign visual stunts
Product announcement campaigns and brand activation content use visual impossibility to generate attention at the awareness stage of a campaign. A product melting — a sneaker, a beverage can, a logo mark — communicates premium material quality through contrast (if it melts, it must be solid; if it dissolves beautifully, the disintegration itself is aesthetic). An AI melt effect on a product hero shot generates a social-first video asset that extends shelf life past a static photograph with minimal production overhead. The same effect applied to a brand logo produces a campaign-specific version of the brand mark for time-limited use without altering the permanent mark.
Concept art and creative direction presentation
Art directors and concept artists presenting thematic directions for film, game, or editorial projects use mood imagery to communicate visual language before production begins. An AI melt effect applied to reference photography generates a tangible demonstration of a specific surreal or horror aesthetic treatment rather than a verbal description. Presenting a melting character portrait alongside a clean version communicates the transformation logic to a creative team in seconds. The generated clip is a working visual artifact — usable in a pitch deck, a mood reel, or a client-facing presentation — not a placeholder for an effect to be created in post-production.
Technical specifications
| Underlying model | Kling 2.1 (by Kuaishou, accessed via Kie) |
|---|---|
| Generation type | Image-to-video (source image upload required) |
| Input | Any still image (JPG / PNG / WebP) |
| Max resolution | 1080p (1920×1080) — native, not upscaled |
| Frame rate | 24 fps |
| Duration | 5 seconds or 10 seconds per generation |
| Aspect ratios | 16:9 · 9:16 · 1:1 |
| Generation time | ~90 s (standard quality) · 3–4 min (1080p) |
| Output format | MP4 (H.264) — video only, no audio |
| Credits — 5 s image-to-video | 12 credits |
| Credits — 10 s image-to-video | 24 credits |
| Storage | 14 days (Free / Credit Pack) · 90 days (Pro) |
| Commercial license | Pro plan |
| Last verified | Kling 2.1 via Kie — June 2026 |
Frequently asked questions
What is the AI melt effect?
The AI melt effect is an image-to-video AI generation that animates a still image into a surreal melting sequence. You upload a source image — a portrait, product shot, graphic, or any photograph — write a prompt describing the melt style, and Kling 2.1 generates the dissolution animation as a downloadable MP4. The effect produces material-aware melting behavior that reflects the visual content of the source image rather than applying a uniform distortion across the frame.
What types of images produce the best melt effect output?
Images with clear edges, distinct subjects, and strong contrast between regions produce the most readable melt sequences. Portrait photos with a clean background give the model a defined face structure to dissolve from. Product shots with hard edges melt in ways that make the form collapse legible. High-contrast graphic designs, including text on a background, generate crisp initial drip formation before dissolution. Very soft or blurred input images can reduce definition in the melt flow since the model has less edge information to work with during animation generation.
How do I control the speed and direction of the melt?
Describe the melt characteristics in the text prompt. 'Slow, viscous dripping from the top edge' produces a measured, deliberate melt. 'Rapid dissolving, fluid and accelerating' generates a faster dissolution sequence. Directionality defaults to top-to-bottom following scene gravity, but you can specify 'melting from the center outward' or 'dissolving from the left edge' to shift the focal point of the collapse. Material type language in the prompt — 'like wax,' 'like paint running,' 'like glass softening' — influences the viscosity and pooling behavior of the generated melt.
Can I use the AI melt effect on copyrighted images?
Use only images you own the rights to or have appropriate licensing for. This includes your own photography, licensed stock images, original illustrations, or visual assets you hold commercial rights to. Generating AI video transformations of images you do not own — copyrighted photography, third-party brand assets, or identifiable likenesses of real people without consent — may infringe on intellectual property rights and violates Polyfaced's terms of service. For branded or commercial output, Pro plan is recommended as it includes the commercial license covering publication rights.
What happens if an AI melt generation fails?
Credits frozen for a failed generation — from upstream processing timeout, content policy rejection at the Kie or Polyfaced moderation layer, or any other cause — are automatically returned to your account balance within seconds. You are not charged for failed or rejected generations. No manual dispute or support request is needed. The refund covers both the Polyfaced moderation pass and the provider-level Kie processing layer.
The AI melt effect is available starting with the 5-credit sign-up grant — enough for one test image-to-video generation at standard quality to evaluate the output before committing. The Pro plan at $29.9 per month provides 800 credits, 1080p native output, 90-day R2 storage with shareable URLs, and a commercial license covering campaign use, client deliverables, and published content. Credit Packs at $4.99 for 100 credits offer pay-per-use access without a monthly subscription. See the pricing page for the full tier comparison.