A video remake or an original rap?
Understand what your portraits change, what the reference supplies and why a familiar performance is different from a newly written rap.
01A rap video remake begins with an existing performance
A rap video remake starts with a reference performance that already has a scene, actions, rhythm and camera work. The creative task is to bring a different person's appearance into that performance. A portrait tells the model who the replacement character should resemble; the template tells it which performance to follow. That is a different starting point from writing an original rap song with a new topic, lyrics, beat and vocal delivery. Both can lead to a short music video, but the required inputs and the meaning of a successful result are different. In the Bulin 47 template, one portrait supplies the appearance reference for the lead in a red-lit street freestyle. The intended result follows the existing scene and performance while changing that main character. You do not need to recreate the street in your photo or describe each gesture. You also do not get a new verse simply by uploading someone whose birthday is coming up. The fixed performance remains the starting point. Watch the scene reference and ask what you want to keep from it: the pacing, setting, movements or familiar moment. If those are the reason you chose it, a remake fits the idea. If you need the performer to say a name, describe a product or tell a new story, check whether the tool actually provides a lyric or topic workflow. The current fixed templates do not offer that editor. A title such as freestyle describes the reference's style; it does not mean every generated clip contains newly improvised lyrics. Understanding this before generation makes comparisons more useful. Judge a remake by how well it carries the intended identity into the reference performance, while judging an original song also involves the writing, music and vocal choices. Neither label alone proves that the result will have a perfect face or synchronized mouth movement.
02Portraits recast the roles; they do not redesign the scene
In a reference-based template, portrait uploads provide appearance information rather than a complete replacement video. The model must interpret a still person during movement, changes in camera angle and speaking. A clear face helps communicate the identity, but an instruction to preserve every gesture does not prove that every generated frame will match. Review the result as a moving sequence. A convincing first frame can still be followed by a changed face, an awkward hand or clothing that shifts later in the clip. Hotel Lobby makes the distinction especially clear because it has two roles. Photo 1 takes the left performer and photo 2 takes the right in an orange studio with one hanging microphone. The name refers to the performance; the current reference is not a hotel lobby scene. Two separate portraits recast those performers while the reference guides their actions, speaking turns and camera timing. The pictures can have different backgrounds because those rooms are not intended to become the stage. Keep the role order consistent and check whether each face remains attached to the right performer when they turn or overlap. A rap video remake is therefore more than choosing a flattering still image. The intended likeness needs to work through the whole performance, and the environment needs to remain coherent around it. The current templates supply the instructions for that task without a free-form scene editor. They are not a way to request an unrelated location, swap every background person or choreograph an entirely new routine. Choose a different reference when a different scene is central to the idea. Also read the example labels: source footage, input portraits and generated results describe different things. A portrait displayed near a scene reference is not automatically a verified before-and-after demonstration. Your own completed task is the evidence for how those particular inputs performed.
03An audio track is different from an identical recording
Sound is another place where a familiar-looking preview can lead to the wrong expectation. An output marked as including sound is intended to contain an audio track. That label by itself does not prove the original recording, the same voices or exact lip synchronization will be preserved. The template's specific instructions explain the intended relationship to the source performance, and the completed file shows what actually happened. The current Bulin 47 and Hotel Lobby templates aim to retain the reference music and rap vocals alongside the replacement characters. This is a generation goal that needs to be checked by listening, not a promise established by a written prompt. A portrait is not a voice recording, so uploading your face does not mean the song will be performed in your own voice. These templates also do not provide a voice-cloning control or a field for rewriting the words. In an original-song workflow, lyrics, music and vocal direction would need their own inputs or creative choices. Do not assume those features exist because an example looks like a music video. Play the reference to understand the intended performance, then listen to your complete result when it is ready. Check the start, transitions and ending as well as the loudest or most recognizable section. Watch whether mouth movement follows the audible performance and whether a voice changes between roles in a duo. A similar beat is not necessarily the same recording, and a recognizable phrase is not evidence that every word or sound is preserved. Before sharing, make sure the music, source material and likenesses are suitable for your intended use. A generated MP4 does not by itself establish permission for every element it contains. Keep the distinction practical: choose the performance you want, understand the audio goal and evaluate the actual file rather than relying on a broad sound label.
04Choose the workflow that matches what you want to change
Decide which part of the idea matters most before choosing a generator. If you want to appear in a recognizable solo street freestyle, open Bulin 47 and prepare one clear portrait. If the idea is two people sharing the existing studio performance, open Hotel Lobby and prepare two separate portraits in the intended left and right order. Those choices change who the reference characters should resemble. They do not change the fixed performance into a newly written story. If new lyrics, a custom name in the verse or an original song are essential, identify a workflow that actually supports those inputs rather than expecting a portrait upload to supply them. Comparing tools by the changes they allow is more useful than comparing only the visual style of a gallery. For a rap video remake, read the photo requirements, watch the reference and review the current output and credit quote. Aspect ratio, duration, quality and sound belong to the selected template; do not assume an option from another scene transfers automatically. Submit after the roles and quote are clear, then follow that task in History. When it completes, check the entire performance for recognizable faces, stable roles, natural movement and appropriate sound. Download the MP4 for a copy on your device and review the saved file before posting it. Present the clip in a way that makes its AI nature clear, especially where viewers might otherwise mistake the performance for a real recording of the people pictured. If you want to try another portrait or another scene, compare the new input and quote as a separate attempt. There is no need to decide that every idea belongs to the same format. A remake is useful when the reference performance is what you want to keep; an original song serves an idea that needs new writing or music. Choose on that basis before spending credits.