Binance / $BNB Hotel Lobby example
The quoted post is associated with a Hotel Lobby clip. Grok reported 125 likes and 276,197 views.
Five follow-ups, six steps, one provenance trail
This collection follows a Grok exchange that starts from Scott｜Monsterblockhk's X post about Higgsfield Genjutsu and a Migos "Hotel Lobby" clip. The chat expands the post into a six-step workflow, then iterates on a source-grounded Turtle model, example videos, and direct MP4 links.
The named examples and engagement counts below are preserved as claims shown in the conversation; they are not independently checked live metrics. The captured exchange begins with the quoted post and Grok's first summary, so an earlier user prompt is not inferred. Open the Grok conversation on X or the source tutorial post.
The practical guide is a Grok expansion of the X post. The six steps below are modeled once in RDF and rendered directly from those entities.
Go to Higgsfield, open Video, then Genjutsu. Grok says the Community page can be used first to find a shared template clip.
Grok recommends a 4–30 second Migos "Hotel Lobby" fragment from Community so the source motion, framing, and music remain available.
Select Object Swap rather than Motion Transfer. Grok describes Object Swap as targeted replacement while retaining more of the original scene.
Add one photo for each person and label the assignments clearly as left/right or Image 1/Image 2.
Use a concise prompt. The source post's wording is: keep Everything the same in the video just replace the faces with the two people in the 2 images provided.
Choose 1080p if available, generate, inspect face consistency, adjust the prompt or references if needed, then download.
Grok’s final draft links three example posts and the tutorial author’s own finished clip. It includes AI-executive pairings featuring Elon Musk and Sam Altman, and Donald Trump and Elon Musk.
Engagement figures and descriptions below are reproduced from Grok’s draft. The conversation did not provide a recoverable capture time, and the figures have not been independently verified.
The quoted post is associated with a Hotel Lobby clip. Grok reported 125 likes and 276,197 views.
Grok presented this as an AI-executive face-swap example.
Grok described this as a high-reach finished example and reported 2,882 likes, 489 reposts, and 760,978 views.
This is the output attached to the source tutorial post. Grok reported 882 likes, 85 reposts, and 205,135 views.
The exchange moves from a brief post summary to a practical guide, then refines the RDF's base IRI and adds examples and direct MP4 media links.
Grok's opening summary says the post describes a two-photo Object Swap for a Migos "Hotel Lobby" clip and aims to preserve the original choreography, camera movement, and music. No opening user prompt is visible in the captured conversation.
User prompt: "Flesh out this howto". Grok expands the summary into prerequisites, six steps, tips, and common issues. The reply adds photo-quality recommendations, a 1080p-if-available suggestion, a consent reminder, and limitations; these are Grok's additions in the conversation.
User prompt: "Redo in RDF-Turtle grounding relative IRIs in the original post url..". Grok drafts Turtle using the Scott post as its base and models the post, quoted post, HowTo, tools, prompts, and steps. This turn establishes the requested source-IRI pattern and starts the RDF iteration.
User prompt asks for two or three popular X examples, especially AI-executive examples. Grok's response names a Binance/BNB clip, an Elon Musk–Sam Altman clip, and a Donald Trump–Elon Musk clip. The prompt's "right now" popularity framing is preserved as a request. The examples and engagement figures in the reply are not independently verified here.
User asks whether the prior Turtle included MP4 links. Grok says it had only X post links, then supplies video.twimg.com MP4 URLs for the examples. The final Turtle draft models those direct URLs as schema:contentUrl and records visible dimensions and durations where supplied.
User prompt: "Okay, make one all inclusive edition so that everything is in one doc". Grok returns a single Turtle draft combining source text, author, workflow, prompts, tips, limitations, consent, examples, metrics, and MP4 links. The collection represents this as the final visible stage of the source conversation, not as independent confirmation of every Grok-generated detail.
Visible as a citation indicator in the conversation; the underlying page list was not inspected in this collection.
The conversation's visible X page is the source for the dialogue and the assistant's claims. Grok's "35 web pages" indicator is recorded as displayed, but the referenced pages are not enumerated here. Popularity labels, engagement counts, tool behavior, account requirements, and durations are conversation-reported rather than independently verified current facts.
Go to Higgsfield, open Video, then Genjutsu. Grok says the Community page can be used first to find a shared template clip.
Grok recommends a 4–30 second Migos "Hotel Lobby" fragment from Community so the source motion, framing, and music remain available.
Select Object Swap rather than Motion Transfer. Grok describes Object Swap as targeted replacement while retaining more of the original scene.
Add one photo for each person and label the assignments clearly as left/right or Image 1/Image 2.
Use a concise prompt. The source post's wording is: keep Everything the same in the video just replace the faces with the two people in the 2 images provided.
Choose 1080p if available, generate, inspect face consistency, adjust the prompt or references if needed, then download.
The source is the visible Grok conversation at the X URL, with Scott｜Monsterblockhk's quoted tutorial post represented as its cited source.
Grok mentions faces that look inconsistent, clips longer than 30 seconds being trimmed, and low-resolution source video producing softer results.
Grok's all-inclusive draft adds a reminder to use only photos of people who have agreed to appear.
It describes using Higgsfield Genjutsu Object Swap to replace performers in a short Migos "Hotel Lobby" clip while aiming to retain the original performance.
Grok's reply lists an account, a short source clip, and two separate reference photos.
The expanded answer contains six steps, from opening Genjutsu through generating, reviewing, and downloading the result.
The conversation quotes: "keep Everything the same in the video just replace the faces with the two people in the 2 images provided."
Grok presents Object Swap as a targeted replacement and Motion Transfer as rebuilding more of a scene. This is an assistant-generated description, not an independently checked product specification.
The draft includes the RyanChi Binance/BNB clip, an Elon Musk–Sam Altman clip, a Donald Trump–Elon Musk clip, and Scott's own attached result.
No. They are values copied from Grok's draft. The conversation capture does not establish when each count was measured.
Yes. Grok's final Turtle draft supplies direct video.twimg.com MP4 links for the four examples, modeled here as named VideoObject resources.
A user-requested category of finished clips; Grok's response names Elon Musk–Sam Altman and Donald Trump–Elon Musk examples.
The video.twimg.com URL stored as schema:contentUrl on a named VideoObject resource.
The edit described in the : substitute the performers' faces while attempting to preserve the source motion and audio.
The video feature or workflow name used in the X post and Grok's expanded how-to.
The Migos source-video example named in the conversation; the answer recommends a short community clip.
A contrasting mode described by Grok as rebuilding more of a scene than a targeted Object Swap.
The replacement mode named in the post and expanded by Grok; the exchange describes it as changing specified performers while retaining more of the source clip.
A still image supplied to guide the identity replacement for one of the two performers.
A likes, reposts, or views value appearing in Grok's draft; the collection does not treat it as a current verified count.
The user's request to make relative RDF identifiers resolve from the original X post URL.
Interactive graph visualization derived from the companion RDF. Click nodes to resolve, drag to explore. Graph data embedded from companion RDF at generation time.
Query this knowledge graph on URIBurner. The editor opens on the canonical SAMPLE entity-type summary (DAV named graph). Pick a recipe, edit freely, then run live or copy.