What the Classic excerpt contains
The three reference frames show the pair from around the waist up, exchanging gestures beside the suspended microphone. Both faces turn during the passage, so choose portraits that remain recognizable beyond a single straight-on pose. The landscape version preserves the original 1920 by 1080 frame. The portrait version places that complete frame inside a 1080 by 1920 orange canvas, with substantial space above and below. It preserves both people rather than cropping tightly around one face; preview that smaller composition before choosing a tall export.
Use the Hotel Lobby Classic template as a baseline
A baseline is one deliberate setup that you can understand after seeing the result. Choose two clear portraits, one screen arrangement, one aspect ratio and a generation tier. Keep a note of those choices. If the result needs work, the note lets you change a specific input rather than guessing what you selected.
For a first attempt, choose a pairing whose appeal comes from the people themselves. Two friends making a familiar appearance together is easier to assess than a concept that depends on a costume transformation or an invented gesture. The site's controls do not provide a custom scene brief, new lyrics or a different song.
Classic is not a quality tier. Standard and Pro are separate generation choices, and neither changes the basic two-person format. Selecting Classic should identify the source performance you want; it should not be read as a promise that the motion is easier, calmer or more accurate than another template. Those comparisons require the actual sources and results.
Check what the preview shows over fifteen seconds
Watch the source preview from start to finish before choosing a face arrangement. At the opening, locate both performers. Through the middle, note when either face turns away or becomes partly obscured. At the end, check that the excerpt remains useful as a short standalone video rather than assuming the opening image represents the whole clip.
A source preview demonstrates the performance being used. It does not demonstrate how your portraits will survive that performance. A clear source face and a recognizable generated face are different pieces of evidence. Keep those questions separate when deciding whether to try the template.
Notice which visual features remain readable at ordinary playback speed. You should be able to follow each subject without pausing constantly. If an important part of your idea needs both people facing forward, verify that condition in the available preview instead of inferring it from a still cover.
Listen too. The template's original soundtrack is retained in this workflow. Your uploaded portraits supply visual identity references, not voice recordings. A source passage that does not suit your intended audience cannot be changed into personalized speech by choosing Classic.
Prepare a pair that gives the model clear evidence
Choose separate photographs in which each person can be identified without zooming far into a large scene. Prefer clear eyes, a visible mouth and a face that is not blocked by a hand, phone or dark reflection. You do not need identical poses, but dramatic differences in clarity make the pair harder to evaluate fairly.
A heavily processed selfie may look polished while carrying less reliable facial detail than an ordinary photograph. Compare the reference to how the person looks to people who know them. Keep characteristic features rather than smoothing them away. Do not enlarge a tiny face and assume the newly sharpened pixels restore information that was absent.
Leave enough of the head visible to understand its shape. Cropping tightly through hair, ears or chin can remove clues that distinguish one person from another. At the same time, avoid uploading a broad crowd scene and expecting the generator to infer which person you intended. This workflow asks for one person in each input.
Ask both people to approve the photographs and the intended audience. Their agreement to share a portrait does not automatically tell you whether they want an animated performance shared publicly. Settle that practical question before generating so the finished file has a clear destination.
Map photographs to screen positions
Use the left and right slots as a visual map. Say the arrangement out loud in simple terms: this photo belongs on the left, that photo belongs on the right. Do not rely on the order of names in a caption or on assumptions about which artist owns a musical passage.
Inspect the arrangement after using any swap control. A swap changes your intended mapping; it does not guarantee that the model will never confuse identities during movement. In the completed video, compare both subjects after turns or moments of overlap, where a convincing opening frame alone would not reveal a later issue.
If the source preview gives one position substantially clearer face visibility during the moment you care about, you can take that into account. That is a choice based on the visible excerpt, not a general rule about the left side. Different source versions must be judged independently.
Compare portrait and landscape as two compositions
The 9:16 version is intended for a tall frame; the 16:9 version provides a wide frame. Inspect the actual Classic preview for the version you plan to use. A ratio label cannot tell you whether hands, hair or the space between the performers remain visible during the entire passage.
On a narrow screen, a wide video may make each face smaller. A tall frame may provide larger faces while offering less horizontal room for movement. Decide which compromise serves your idea. A post about a recognizable pair may prioritize faces; a post about their visible interaction may need more space around them.
Do not judge the frame only from the cover. Pause near the widest visible movement and check both outer edges. Then play the excerpt normally to see whether the composition feels stable. If either person repeatedly leaves the useful area, choose the version that keeps the relevant action readable.
Cropping a completed landscape video into portrait later is an editorial choice with possible losses. It is not automatically equivalent to generating from the dedicated portrait source. Plan the destination early if preserving both people in a tall post is essential.
Decide what a successful first take means
Set a small acceptance list before you spend credits. For a baseline, the two people should stay recognizable during normal playback, keep the intended positions and fit the chosen frame. The soundtrack should play without an unexpected gap, and the mouth movement should not distract from the performance. These are review criteria, not guarantees.
Standard uses 300 credits. Pro uses 400. A template selection by itself should only prepare a choice; generation requires the explicit Generate action. If a sign-in or credit purchase is needed after that action, follow the existing flow without creating several separate requests because a screen appears to be waiting.
Review a completed take once with sound and once while watching the faces closely. Note any problem with a timestamp and a short description. A useful note is left face changes after turning, not the result feels wrong. Specific observations help you decide whether to replace a portrait, change placement or try another available source.
Keep the result or change one decision
If the clip meets the acceptance list, download it and prepare the caption around the relationship between the two people. There is no need to regenerate simply because another template exists. A usable baseline is already a finished short video.
If it misses the list, identify the smallest meaningful change. A blurry reference suggests choosing another photograph. A composition problem suggests comparing frames. A passage that does not fit your idea suggests another source. Changing all three at once makes the next result harder to interpret and may spend credits without teaching you much.
When the Hotel Lobby Classic template preview is ready, use that exact version to confirm your setup. Until then, keep your two approved portraits and creative brief ready, and avoid treating this preparation guide as evidence of a tested Classic result.