Knife blocks: first-last sheathed to drawn, same handle steel
Two photos of the same SKU. Interpolate. Do not let the blade count change.
Two photos of the same SKU. Interpolate. Do not let the blade count change.
A knife-block listing is not a chef talking over a board. The offer is a state change: the set is sheathed, then the set is drawn, and every handle is still the steel you sell. That is a bounded motion problem. You already own both stills. Do not prompt a model to invent the block.
Create a first-and-last-frame transition video is the named agent job: two photos in, one transition out. The agent passes them as first and last frame and picks a first-last-capable model. Cost is priced per model, shown before you confirm. This is not the plain "turn one photo into a video" capability.
Sheathed is frame A; drawn is frame B
Photograph the loaded block where it sits — same lens height, same counter, same light family. That is the first frame. Photograph the same block with the knives drawn to the positions you actually sell, or the same set fanned in front of the empty slots if that is the listing. That is the last frame.
First-last frame takes those two stills and generates the motion between them. Ordinary image-to-video knows where it begins and improvises the rest. First-last has to land. The arc of the shot is decided by you.
The two frames have to be plausibly connectable in the time available. Same block, same slot count, same handle steel, same camera. A close-up of a chef's knife morphing into a wide of an eight-piece set is a warp, not a draw. Keep the block. Count the blades in both stills before you spend a second. If frame A is a five-slot oak and frame B grew a steel and a pairing knife, you are interpolating two SKUs.
First-last frame is the product-turn model is the same contract on a pack shot: pin both ends so the object survives. Here the object is the set. Blade count is the label.
Interpolation is the listing
Flux 3 First Last Frame to Video is the catalog row for that interpolation: start frame and end frame, native audio, 720p or 1080p, durations from 5 seconds through 20. The listed rate is 9 credits per second. Audio is on — write the mix (a quiet wood, a faint steel tick) or write silence. A silent brief still gets a soundtrack you did not choose.
Prompt the travel, not the objects. The stills already contain the handles and the slots. Name the mechanism: knives draw from the block, camera locked, no new steel, slot count unchanged. If the endpoints share geometry, the morph reads as a draw. If they do not, you get a mushy crossfade you could have made in an editor without a video bill.
Duration should match the change, not a Reels quota. A five-piece set does not need 20 seconds of travel. A 20-second morph is a different bill at 9 credits per second, and the middle is where a generator cheats — an extra blade, a handle that swaps wood for a cousin steel, a slot that heals itself. Inspect the midpoint. Ends are held by construction. If the middle grows a ninth knife, reroll the interpolation. Do not regenerate both stills "to match the motion." The stills are the contract.
The photographs are the product. The prompt is the camera.
What this is not
It is not a talking chef. It is not a text-to-video "walnut block, knives fly into frame" paragraph that rebuilds the set. It is not a single-image wiggle that leaves the knives sheathed.
Do not use a last frame that is a different SKU. A seven-piece hero into a five-piece sale plate will connect. The connection is the artefact — blades appearing or vanishing mid-draw. A loop (same still in both slots) is ambient motion on one plate. This offer needs two states: sheathed, then drawn.
AI video for butcher shops is the nearby audience: the block as theatre, a primal breakdown. That playbook is a shop that bills cuts. It is not a CPG knife-block listing. Do not borrow a generated butcher as a host. Steel grade is overlay after picture lock, not a mouth in the interpolation.
Count the blades, then generate once
Ask: do I know the start picture and the end picture, and is the point to travel between them without inventing a new set?
Yes → first-last. Attach sheathed. Attach drawn. Same handle steel. Same slot count. Generate once.
I only have the loaded still → that is ordinary image-to-video. The model will guess how many knives leave the block.
I do not have the drawn photo yet → shoot it. A generated "after" is a different set.
Atmosphere on the empty block after the draw can be a later pass of frame B. It is not a substitute for the interpolation.
FAQ
Can I animate one photo of the loaded block and hope the knives draw?
You can. The ending is then the model's guess. A knife-block listing is the drawn set, same steel, same count. Pin the drawn plate as the last frame or you are not selling the SKU.
Why Flux 3 instead of a talking-chef model?
Because the product is interpolation between two approved stills, not a mouth. Flux 3 first-last is 5–20 seconds, 720p/1080p, native audio, 9 credits a second. A presenter row will talk over a block it invented.
Do the two photos have to be the same block?
They have to share geometry and count. Same slots, same handle steel, same lens family, same lighting neighbourhood. A noon oak into a night walnut is a time-lapse you should prompt as one. A five-piece into an eight-piece is a morph.
Is this the same as uploading one image to image-to-video?
No. One image is open-ended motion. Two images is a bounded landing. The agent notes that this capability needs first frame and last frame. Missing the drawn set is a different job.