The AI Squish Trend, Explained (No Preset? Here's the Fix)

The AI Squish effect takes a photo of almost anything — a person, a pet, a plush toy, a product — and animates it compressing down like memory foam or mochi, then slowly bouncing back into shape. It's a short video, not a still image, and it's been all over TikTok and Instagram in 2026 for exactly that oddly-satisfying reason. Flovaly doesn't have a dedicated one-click Squish preset the way it has the Age Ripple or Long Hair Magic— but the Studio's Video tab can produce the same kind of motion from 25 credits, and this article is honest about how that workflow differs from a dedicated squish generator.

What the trend actually is

“AI Squish It” videos follow the same three-beat motion every time: compress, bounce, recover. Creators upload a photo — a squishy toy, a car, their own face, a pet — and an AI video model simulates soft-body physics on it, so the subject deforms under invisible pressure and springs back like elastic or a stress ball. Search results for the trend are split between dedicated one-click generators built specifically for this look, and creators pasting a squish-motion prompt into a general text-to-video or image-to-video model like Veo 3, Kling, Sora 2, or Hailuo. CapCut templates cover the rest. It travels fast because the physics reads instantly, even muted, and works on almost any subject.

Why this isn't a one-click preset on Flovaly

A grep of Flovaly's full Viral Centre scene list — all 32 presets across Trending, Transforms, Effects, and Reactions — turns up nothing for squish, squash, mochi, or plush distortion. If you open the Viral Centre looking for a scene to drop a photo into, there isn't one. The closest real workflow uses the general Studio instead, on the Video tab rather than the Image tab — squish is motion on the whole clip, not a new still frame, so the compose-then-animate trick that covers other no-preset trends like the Mini Me trend doesn't apply here.

How to make a squish-style clip on Flovaly

1. Open Studio in the dashboard sidebar and switch to the Video tab, then select prompt mode.
2. Pick any saved character — a custom actor or a premade avatar. The composer requires one to be selected, but it doesn't have to match your subject; the next step overrides it.
3. Use Add reference imageto upload the actual photo you want animated — your squishy toy, pet, product, or selfie. That upload, not the character's own photo, becomes the video's starting frame.
4. Write a motion prompt describing the squish: for example, “the subject compresses and flattens like soft memory foam under gentle pressure, then slowly bounces back into its original shape.”
5. Pick a base video model that supports image-to-video — Seedance 2.0 Pro (the default) or Kling 3.0, the cheaper option. Set duration and resolution, check the credit cost shown above the Generate button, then generate.

The one honest limitation

Two things are worth knowing before you try this. First, the required-character step is a genuine quirk of the composer, not a creative choice — you have to pick an existing character to satisfy the form, even though your uploaded photo is what actually gets animated. Second, and more importantly: Seedance 2.0 Pro, Kling 3.0, and Veo 3.1 Fast are general-purpose image-to-video models, not models trained specifically on squish physics the way a dedicated one-click squish generator is. The motion prompt gets you close, but the compression-and-bounce result can vary between attempts — expect to regenerate once or twice to land a clip that reads as cleanly “squishy” as the purpose-built tools. Sora 2 Pro, Flovaly's other video model, can't be used for this at all — it's text-to-video only and won't accept the uploaded photo as a starting frame.

Getting a convincing result

• Name the material.“Like memory foam,” “like mochi,” or “like a stress ball” gives the model a physical reference; a bare “make it squish” prompt leaves the texture and speed of the deformation to chance.
• Describe the full beat, not just the squish. Compress, hold, bounce back — spelling out the recovery half of the motion helps stop the clip from ending mid-compression.
• Use a clean, well-lit source photo. A subject shot against a simple background gives the model less to reconstruct around the deformation.
• Start at 720p and the shortest duration.At 25 credits for a 5-second Kling 3.0 pass, it's cheaper to confirm the motion looks right before paying for a longer or 1080p re-render.

What it costs

Pricing follows Flovaly's standard video-model rates, not a fixed Viral Centre price, since this runs through the general Studio rather than a preset scene. A 5-second clip at 720p is 25 credits on Kling 3.0, 30 credits on the default Seedance 2.0 Pro, or 45 credits on Veo 3.1 Fast. Switching to 1080p multiplies that by 1.5 — 38, 45, and 68 credits respectively — and longer durations (10 or 15 seconds) scale up from there. That's meaningfully more than a Viral Centre preset like the Age Ripple(from 5 credits), because there's no cheaper preset pipeline underneath it. See pricing for how credits map to each plan, or the full cost breakdown for how this compares to every other generation type on Flovaly.

Where this fits

If all you want is a single squish clip with the least effort, a tool built specifically for this trend will get you a more reliably squishy result in fewer attempts — that's the real trade-off of using a general-purpose video model instead of a purpose-built one. Flovaly's advantage is that the same Studio and the same saved character you'd use for this also make the rest of the Viral Centre line-up, talking-actor ads, and product UGC — so it's worth it if you're already making other content, not if a single trend clip is all you're after. It's a $1 trial to see how close the result gets to your own photo. The same Studio workaround also applies to two other physics-driven trends without a preset: the Melting effect and the Bird Dance trend.

FAQ