You get a short product clip made from one photo that finishes on a clean logo card, built on one 8frame board: an image node for the card, a video node for the product shot, and either an end frame, a Montage node or the Video Editor to join the two. The limit people do not expect: Montage (the Merge Videos node) joins the picture only, so the merged file has no sound, and any voice, music or generated audio has to go back on in the Video Editor.
There are three honest ways to end on a logo, and they differ in what the model is allowed to touch. The safest keeps your real logo file away from the model entirely.
TL;DR
| Step | Node or model on 8frame | Fixed price | Needs as input |
|---|---|---|---|
| Product shot | Kling v3 Standard | 86 credits for 5 s with sound, 57 silent | A start image (Kling v3 is image-to-video only) |
| Product shot that lands on the card | Kling 2.5 Turbo Pro | 48 credits for 5 s, 95 for 10 s | Start image and end frame |
| Product shot at 1080p (routes B and C) | Kling 2.6 Pro | 95 credits for 5 s with sound, 47 silent | A start image; no end frame on the canvas |
| Card background or scene still | Nano Banana 2 | 11 credits per image at 1K | Prompt, optional references |
| Join product clip and card clip | Montage (Merge Videos) | 10 credits per run | 2 to 10 clips; no sound in the output |
| Logo, audio and text | Video Editor timeline | no generation | Your logo file, audio, script |
The board, node by node
This is a blank board, or the clip maker plus a few extra nodes. If the product first needs a scene, PRODUCT PLACEMENT on the workflows page is the place to start; the end card comes after it.
1. The product photo
Upload the real product photo. If you only have a cut-out, place it into a scene first with an image node. Keep the product in the frame you will deliver: 9:16 for Reels and TikTok, 16:9 for YouTube and the website.
2. The product shot
Connect the photo to a video node. Kling v3 Standard is what the Kling tool opens on, 86 credits for 5 seconds with sound. It needs a start image, which you have. Keep the camera move small: one push-in or one slow orbit. A short, calm move gives the model less room to change the product before the cut.
3. The end card as a still
Design the card as an image, not as video. You have two options:
- Your designer's card. Export the end card from your design tool (logo, tagline, URL) as a PNG at the delivery ratio. This is the version where nothing about the logo is generated.
- A generated background, your logo on top. Generate only the background on an image node (a colour field, the product's surface, a soft gradient in brand colours) and add the real logo file later in the Video Editor. Image models do not spell reliably, so a generated wordmark or tagline drifts; the same reason a label changes is covered in why AI changes a product label.
4. Join the two
This is the choice that matters. The next section compares the three routes.
5. Finish in the Video Editor
The Video Editor is a separate mode with a timeline, video, text, caption and sound layers. Put the joined clip on the timeline, add the logo file as a layer over the last seconds if it is not already in the card, add a voice-over or music on a sound layer, and set the type for the URL or offer.
Three ways to end on the logo
Route A: the model lands on the card (end frame)
Give the product photo as the start image and the card as the end frame on a model that takes both (which ones is on first and last frame in AI video; the Kling tool's default, Kling v3, does not), and the model generates the move into the card. The catch is that it also generates the frames where the logo forms, so text on the card can warp for a moment before the last frame. Use this route only for a mostly graphic card, and add any words in the Video Editor afterwards.
Route B: two clips joined in Montage
Animate the card on its own (a slow zoom on the card still is enough) and join it to the product shot with the Montage node. Montage takes 2 to 10 clips, uses at most 15 seconds of each (1 to 15 s, default 5), offers cinematic, hyper and clean transition styles, and costs 10 credits per run.
The catch is the one in the first paragraph: Montage's output has no audio track. If your product shot came out of Kling v3 or Kling 2.6 Pro with sound, that sound is gone in the merged file. Montage is fine for a silent cut; for anything with sound, use route C.
Route C: a still on the Video Editor timeline
Put the product clip on the timeline, then the end card image after it for two or three seconds. Nothing on the card is generated and nothing is re-rendered by a model, so the logo is your file, pixel for pixel. The product clip keeps its own sound, and you add music under both.
For most product ads this is the route to start with. Route A is the one to try when you want the product to move into the card in one continuous shot.
Which model for which step
| Model | End frame on the canvas | Length and resolution | Sound | Price |
|---|---|---|---|---|
| Kling v3 Standard | No | 3 to 15 s, 720p | Toggle, on by default | 86 for 5 s with sound, 57 silent, 171 for 10 s |
| Kling 2.5 Turbo Pro | Yes | 5 or 10 s, 1080p | Files had no audio track in our sample | 48 for 5 s, 95 for 10 s |
| Kling 2.1 Pro | Yes | 5 or 10 s | Not stated here | 67 for 5 s, 133 for 10 s |
| Kling O3 | Yes | 3 to 15 s, 720p | Off by default | 57 silent, 76 with sound for 5 s |
| Kling 2.6 Pro | No | 1080p in our measurement | Toggle | 95 for 5 s with sound, 47 silent |
| Seedance 2.5 | Yes (start image required) | 4 to 30 s, up to 720p | Audio included | 157 for 5 s at 720p |
The table is about inputs, not quality. We have not compared how cleanly each model resolves into a logo; test the same start and end pair on two models before you commit a campaign to one.
The sound column comes from our own file check: Kling 2.5 Turbo Pro files in our sample carried no audio track, while Seedance 2.5 files did. If you need the clip silent anyway (route B), a model without sound costs less for the same seconds.
Where it breaks
- Text in generated frames. Any words the model draws can misspell. Keep words off generated pixels and add them as text or image layers.
- Montage removes sound. Said twice on purpose, because it is the most common surprise. Join talking or lip sync clips on the Video Editor timeline, never in Montage.
- No 4K video and no video upscaler. No video model on 8frame outputs 4K, Seedance tops out at 720p, and the upscalers (Topaz, BRIA, Clarity) work on images only. You can upscale the card still before it goes on the timeline; you cannot upscale the clip.
- The end frame is a target, not a lock. On route A, check the final frames at full size: if the logo is soft or shifted in the last frame, cut to the real card still for the final second in the Video Editor.
- Ratio mismatch. Make the card at the same ratio as the clip. A 16:9 card on a 9:16 timeline is either letterboxed or cropped.
Platform rules for the last seconds
Two first-party rules change how you design the card.
YouTube end screens. YouTube lets you add end screen elements to the last 5 to 20 seconds of a video, but only if the video is "at least 25 seconds long", and you can add "up to four elements" on a standard 16:9 video. If your product video goes to YouTube at 25 seconds or longer, those clickable elements sit over your card, so leave room: a centred logo with empty space around it, not a card filled edge to edge.
Instagram Reels ads. Meta keeps the bottom of a Reel for the caption and call to action, so a logo placed low on a vertical card sits under them. Put the mark in the upper-middle band.
FAQ
Can AI generate my logo on the end card? It can draw something that looks like a logo, but image and video models do not reproduce lettering reliably. Use your real logo file as a layer in the Video Editor, and let the model generate only the background or the motion.
Why does my merged video have no sound? Montage (Merge Videos) outputs picture only: the merged file has no audio track, so the clips' own sound is dropped. Join clips with sound on the Video Editor timeline instead.
Which models can end on a frame I choose? Several Kling models and Seedance (with a start image) take an end frame on the canvas; the full list is on first and last frame in AI video. Kling v3, the Kling tool's default, cannot.
How long should the end card be? Long enough to read the logo and one line, usually two to three seconds. On YouTube, if the video is 25 seconds or longer and you use end screens, plan for the 5 to 20 seconds they can cover.
Can I make the whole thing from one product photo? Yes. The photo is the start image for the product shot, and the end card is either your own design file or a generated background with your logo added on the timeline.
Sources
- 8frame prices, defaults and limits: the 8frame app, checked 2026-11-09 (Kling v3 Standard, Kling 2.6 Pro, Kling 2.5 Turbo Pro, Kling 2.1 Pro, Kling O3, Seedance 2.5, Nano Banana 2, Montage)
- End-frame support per model on the canvas, Montage limits and the missing audio track in merged files: the 8frame app, checked 2026-11-09
- Audio track in output files by model: 8frame read-only check of generated files, 120 days to 2026-10-02 (shares only)
- YouTube end screens, last 5 to 20 seconds, "at least 25 seconds long", "up to four elements": YouTube Help, Add end screens to videos, checked 2026-11-09
Try it without signing up: the 8frame canvas opens on a real board. The canvas is free and unlimited; generation is paid from $19/month.