A photo of a closed gift box can start two very different shots. Want the camera to drift a little closer? One image usually does the job. Need the lid to open and reveal something specific? Then an open-box photo gives the model a clear ending to aim for. So the real question is whether the final image matters enough to show it in advance.
Wan 3.0 AI Video Generator asks for a first frame in its Image to Video mode, and you can add a last frame if you have one. The first frame sets how the shot opens. The last frame gives the model a picture to work toward. That extra image is worth preparing when the final view is part of the story or has to match the next shot.
When one photo is enough
Say you’re making a short teaser for a gift guide. You don’t want to show what’s inside the box yet; you just want a little movement behind a title. A clear shot of the closed box gives you a starting point. Leave the last frame empty and let the prompt describe the camera move:
The camera slowly moves closer to the closed gift box. The lid stays shut. Keep the whole box in view and the lighting steady.
There’s no particular final composition to match here. Watch the preview, find the moment the box fills enough of the screen, and trim the clip there. If a title will sit beside the box, check that the camera move leaves room for it. And add that title in your editor rather than asking the model to put text into the scene.
When the final picture matters
Now picture the clip introducing what’s inside the box. The lid opening stops being a bit of background movement. Your audience needs a clear view of the inside before the video cuts away. A prompt can ask for an open lid, sure. But a second photo shows how far it should open and what the camera should be able to see.
Take one picture with the box closed and another with it open. Leave the camera where it is and keep the box in the same spot. Shooting on a phone? Prop it up instead of trying to match the first shot by hand. Check that the open lid fits inside the frame before you take the second picture.
The two photos don’t have to look identical. The lid should move, and the inside may catch more light once it’s open. Try to keep unrelated things steady, though. If the second picture is also much closer, or the box has wandered across the table, the model has more changes to reconcile than the one you actually want.
Describe what happens between the two pictures
In the Wan 3.0 image-to-video tool, drop the closed-box picture into the first-frame field and the open-box picture into the optional last-frame field. The pictures mark where the action starts and ends. Save the prompt for the part they can’t show: how the lid moves.
Keep the camera still. The lid lifts slowly on its hinge until the box is open. The box stays in the same place on the table. Hold on the open box briefly.
A last frame can guide the ending, but it won’t tell you whether the movement in the middle looks right. So watch the full clip. Does the lid look attached as it rises? Does the box slide when it should stay put? Is the inside on screen long enough to register? If the lid seems to jump open, try simplifying the prompt or bringing the two photos closer in framing before you pile on more instructions.
Still unsure whether the extra photo helps? Run a quick side-by-side, the same fixed-input habit teams rely on to test AI video models fairly. Keep the same opening photo, prompt and settings, and add the open-box photo only to the second version. Watch both at the length you’ll actually use. If the second version makes the contents easier to see without the lid jumping or the box sliding, the extra photo earned its place.
Use the next shot to make the decision
Look at what you plan to put right after the clip. If you storyboard your AI videos before generating anything, that answer is already sitting in your shot list. Cutting to a close-up of what’s inside? An open-box last frame can help set up that cut, and you can compare the end of the generated clip with the start of the close-up in your editor to pick a cut point that makes sense. Cutting to a title card, with the box meant to stay closed? Then an exact ending may not matter at all.
This helps when you’re working from someone else’s photos, too. If all you have is a closed-box image and the contents must be accurate, get a real open-box photo or cut straight to a separate product image. A generated reveal isn’t a reliable way to show details that were never visible in the source material.
Before you open Wan 3.0, finish this sentence: “At the end of the shot, the viewer should see ___.” If the answer is simply “the same box a little closer,” start with one image. If it’s “the lid open, with the inside clearly visible,” take the second photo. Then check the whole clip, not just the ending. The reveal only works if viewers can follow how the box got there.
Related: Image to Video AI: What Does a Usable Clip Really Cost?
