DEV Community: john smith

The Portrait Was Correct. Why Did the Reference Video Replace the Person?

john smith — Sun, 20 Sep 2026 08:01:05 +0000

You give an AI video tool a clear portrait of the person you want to see. Then you add a short reference video because you like its movement or camera angle. The clip starts well, but halfway through, the face, clothing, or body shape begins drifting back toward the person in the original video.

It can look as if the main image was ignored. More often, two references are being asked to answer the same question.

A portrait says, “This is the person.” A video can quietly say the same thing through its actor, clothes, posture, framing, and movement. When both files show a visible person, the model has to decide which identity should lead. A prompt such as “replace the person with my character” may not be enough to resolve that conflict.

The Reference Video Is Carrying Identity, Not Just Motion

Imagine you have a portrait of a woman in a mustard-yellow coat. You also have a three-second subway clip of someone else walking past the camera. You want the woman from the portrait to perform the walking motion from the subway clip.

The intention is simple: keep the woman, use the motion.

But the subway clip contains much more than movement. It also includes another person’s face, hairstyle, coat shape, proportions, posture, lighting, and distance from the camera. It is a complete visual scene with its own identity cues.

That is why a result may shift back and forth. The still image and the video clip are supplying competing answers about who should appear on screen.

Give Each Input One Job

Before generating, write one short sentence for each reference file.

For example:

Portrait image: defines the character’s face, hairstyle, yellow coat, and overall appearance.
Subway clip: provides only the walking pace, camera distance, and direction of movement.
Text prompt: defines the new setting and the single action.
Ending frame, if the chosen generator accepts one: defines where the character should stand at the end of the shot.

This will not make every system separate those roles perfectly. But it can reveal a problem early: a motion clip with a clear, visible actor may contain too much identity information to work as a motion-only reference.

When that happens, a longer prompt is usually not the first thing to try. Start with a simpler test instead.

Test Identity Before Adding Motion

Use the portrait image and one plain-text action. Keep the camera still, or give it only one gentle move.

For the subway example, the test could be:

The woman in the mustard-yellow coat walks forward for three steps in a bright subway station. Medium shot. The camera stays at the same distance. No cut and no other person enters the frame.

If the character changes in this simple version, check the portrait, the scene complexity, or the action before adding a reference video.

If the character stays stable, add the motion video in a second version. Compare the first frame, the middle of the movement, and the final frame. The moment when the character begins to drift is more useful than a general feeling that the full clip is “wrong.”

Use Motion References Without Visible Faces

Sometimes a reference video is useful only because of its camera motion. In that case, a clip containing another visible actor may be the wrong source material.

A simpler motion reference could show:

an empty hallway with a camera moving forward;
a close shot of footsteps without showing the upper body;
a silhouette with no recognizable face or clothing;
a simple object moving across the frame.

Reducing unrelated identity information in a motion reference may make it easier for the model to follow the intended character reference. The most useful clip is not always the most detailed-looking one. It is the one that supplies the missing movement instruction without introducing a second subject.

Do Not Solve Everything in One Generation

A short test works best when it answers one question at a time:

Can the portrait hold the intended identity during a simple action?
Can the scene change without changing the person?
Does the camera movement still work after the motion reference is added?
Does the final frame match the planned result?

If several things fail at once, remove one input instead of adding more instructions. A crowded prompt, two visible people, a scene change, and a moving camera can hide the real source of the conflict.

If you are preparing this kind of test in Vidu Q3, treat the first result as a small diagnostic draft rather than a way to make conflicting references automatically agree. Give each reference one clear job, test the smallest useful motion first, and check whether the intended person stays stable before adding more controls.

Check the First, Middle, and Last Frame

Pause the clip at the beginning, middle, and end. Ask:

Is the intended person still recognizable?
Did one action happen clearly from start to finish?
Did the motion reference add useful movement, or did it pull in the original actor’s appearance?
Is there one input you can remove without losing the core idea?

If the answer to the last question is yes, remove it for the next attempt.

A reference image and a reference video can work together, but they should not both be asked to define the same person. Decide which input owns identity, give the other one a narrower role, and test the smallest version of the shot before building anything more complicated.

When a motion reference includes another visible person, do you remove it first, crop it down, or keep it and rely on the prompt to separate the roles?

The AI Ad Sold the Mood. It Forgot the Product.

john smith — Fri, 18 Sep 2026 02:18:22 +0000

The coffee ad looks lovely. Morning light falls across a kitchen. Someone makes a cup, pauses, and enjoys a quiet moment. Then the video ends and you realize you cannot name the coffee brand—or even remember seeing the bag clearly.
That is an awkward result for an ad. The footage may be attractive enough to stop a scroll, yet leave the viewer with no idea what was being sold. If that happens, improving the lighting or making the image sharper is unlikely to fix the main problem. The product needs a clearer place in the scene.

Try a three-second, sound-off check

Play the opening without audio and pause after roughly three seconds. Ask someone who has not seen the brief: “What is this selling?” Then look at the same frame at phone size. Can you still spot the coffee bag? Could you read the brand, if reading it matters?
Three seconds is not a rule that every ad must obey. It is a quick way to catch a mistake before you spend time polishing the wrong shots. If the answer is “a cozy morning,” the mood is working. The product is not doing enough yet.
This test also tells you what to change. When the bag is visible but too small, fix the framing. When it is large enough but a hand or kettle keeps covering it, fix the blocking. When the lettering turns into gibberish, do not expect a longer style prompt to repair the brand.

Place the product before dressing the scene

Start with one still frame. Put the bag and the cup where they can both be seen. Face the front of the package toward the camera. Make sure the person’s hand can complete the action without hiding it.
Only then decide how much window light, background detail, film grain, or kitchen clutter to add. Doing this in reverse can produce a beautiful lifestyle image with no comfortable place left for the product.
A useful prompt describes what the viewer needs to see, not just the desired mood:
A coffee bag faces the camera beside a cup. One hand pours coffee once without covering the bag. Keep the camera steady and the package visible throughout.

That is more actionable than “cinematic, warm, authentic, handheld.” Those words may describe the feel you want, but they do not tell the shot what the viewer needs to recognize.
For a real campaign, use packaging images and brand assets you are allowed to use. Check the generated frame against them. If the name must be exact, add approved lettering in the edit rather than relying on generated text. A plausible-looking label that spells the brand incorrectly is still an error.

Give the camera one reason to move

Handheld movement can make a scene feel lived-in. Too much of it makes the viewer follow the shake instead of the coffee. In a short ad, one gentle move is often enough: a slow push toward the table, or a locked-off shot while coffee is poured.
Hold the product clearly in frame for a moment. Then watch the clip again and ask whether the camera move helps someone notice the product or merely shows off motion. If it does the latter, simplify it.
This is not an argument for perfectly still footage. Camera movement just has to serve the ad’s point, especially when the viewer has only a few seconds to understand it.

Show the moment after preparation

Brewing coffee tells us what is happening. It does not necessarily make us imagine drinking it. One small use moment can close that gap: a hand lifts the cup, someone takes a sip, or a cup is passed to a friend.
Pick one action that suits the brand. You do not need a full story arc, three camera angles, and a dramatic ending in a 10-second clip. If the package, the pour, and the first sip cannot fit clearly into one shot, split the idea into a couple of simple shots rather than making every second busy.
For early shot experiments, the independent third-party MiniMax H3 site currently offers 4–15 second video generation from text, frames, or reference material. An authorized product image could help you explore composition and motion there. It does not guarantee readable packaging, stable object details, or a successful ad; check every output and finish the brand-critical parts in editing when needed.

Show it to someone without the brief

Before publishing, play the draft for someone who does not know the concept. Do not supply a title, logo slide, or explanation first. Ask three questions:

What was being sold?
When could you first see it clearly?
Did you see someone use or enjoy it? If they remember the light, music, and kitchen but not the coffee, resist the urge to rebuild the whole visual style. Try a larger product in the opening frame, a calmer camera move, or a clearer use action. Change one thing, then run the same check again. The mood earns the first look. The product still has to survive the second one.