Alvin's Club Studio

Only when the user asks to use their own model (a real person). Paid accounts; costs 10 credits once. Before calling, show the user this one statement and get a clear yes: "I confirm I'm the person in these photos or have their written permission, they're 18 or older and not a public figure, and I agree to use 10 credits (once)." Then call with consent=true (it covers the three consent statements and the charge). image_urls: 1-3 public links to photos of the same person; if the user only has photos in the chat, send them to studio.alvinsclub.ai/app/models to add the model there. Before asking for photos, send the user the example photos at https://studio.alvinsclub.ai/models/photo-guide (what works and what to avoid) and tell them these photo requirements: Photo 1: a close-up of the face and shoulders, facing the camera, eyes open, natural expression. Photo 2: full length, head to feet, standing straight and facing the camera, arms relaxed, in fitted clothes so the body shape shows. Photo 3 (optional): the face from the side, or half length. Only the person who agreed to be your model, alone in every photo: no one else, no mirror selfies. Nothing over the face: no sunglasses, hats, masks or hair across the eyes. Soft, even light (daylight facing a window works); no strong shadows or backlight. A plain, uncluttered background. No filters, beauty retouching or slimming. No text, stickers, logos or watermarks; not a screenshot. Sharp and in focus, at least 1024 x 1024 pixels; JPEG, PNG or WebP, under 4 MB each. Photos are checked (one clearly visible adult, no celebrities, no screenshots); the answer says why if they can't be used.

create_custom_model

Get a link that connects the user's own LinkedIn (needed once, then every 60 days). Give connect_url to the user to open in a browser where they are signed in to Alvin's Club Studio, and allow posting on LinkedIn; then check get_linkedin.

connect_linkedin

Publish a post on the user's own LinkedIn now, with the final text they approved (#words become hashtags). Optionally link_url (a card; title from the page unless link_title) OR media_ids from upload_linkedin_media (one video, or 1-20 photos). For a later or recurring post, use client-side scheduling only if available; this server does not schedule posts.

create_linkedin_post

AI stylist lookbook: the product styled on an AI model in real-world scenes, one photo per look (max 8). Each account gets 5 free sets a day; after that a set costs 1 credit: if the answer says so, tell the user the price and their balance, and only with a clear yes call again with accept_charge=true. Give product_url, or image_urls plus product_name; for an outfit give product_urls: 2-4 product pages worn together. looks: styling briefs such as 'Weekend brunch: white sneakers'; omit for 4 default occasions. Keep a chosen model; otherwise omit model_id and let us choose. Only fetch model previews when asked to choose. Photos take 1-2 minutes: then get_lookbook.

create_lookbook

Create product videos (clothing, accessories, bags, shoes, jewelry, chairs, furniture) after the user confirms the complete recommendation and total credits. Pass product_category (womens_apparel, accessories, bags, footwear, jewelry, seating or furniture) when known; if omitted it is read from the page (the result says which), and when the page doesn't say, nothing is made or charged: ask the user what the product is and call again with it. Default to one 8-second video (3 credits) unless specified. 12 seconds costs +1, 15 +2 each and require a paid account. Do not separately ask for length or model: pick one of our models, or let us choose (omit model_id; for home goods the answer names the model we picked: tell the user once); show model photos only if asked to choose. Pass expected_total_credits and a unique idempotency_key, reused only for the same order. For a new version, read get_video_reuse_options first. Returns immediately: share page_url and actual started/not_started counts. Usually ready in about 10 minutes. Do not promise proactive notifications unless the client supports them. Generated videos are delivered and charged, including edits and home products. Quality warnings do not trigger refunds. use_my_voice=true only when the user asks for their own voice and has set it up at /app/voice (get_account shows has_voice). Voiceover is English unless the user asks for another language: language=es (Spanish), pt-BR (Brazilian Portuguese), de (German), it (Italian), id (Indonesian), ja (Japanese), ko (Korean), zh (Chinese). Spanish & Brazilian Portuguese voiceover: clothing, chairs and furniture for now. The caption and hashtags follow it; same price. Only when the user gives a viral TikTok video to model it on (TikTok links only for now; Instagram and YouTube are not supported yet): pass it as reference_video_url (needs product_url; costs +1 credit per video, include it in the total). We borrow only its structure (pacing, hook, what each shot does) with this product's facts and our model, never its footage, sound, words or creator; passing it means the user confirms they may use that video as a reference (Terms, 'Reference videos'): say so when confirming. replica_mode defaults to swap_product (the video shows another product). Use same_product only when the user says the video already shows THIS product and they want its moves kept; then first get their yes to 'The product in that video is this product, and I have the right to use that video' and pass same_product_confirmed=true. Moves our review never allows are left out and listed on the result. Same-product remakes without duration_seconds are 15 seconds (8 on the free trial; shorter only when the reference's length is known and shorter): quote that price and always pass the confirmed expected_total_credits (it is required then); if the answer says the total changed, confirm the new total. For a product sold in several colors (or styles), pass variant (its name on the product page, e.g. 'Black') only when the user names one; otherwise we pick the one closest to the reference video, else the page's default. Tell the user the color in the answer's product_variant once. For 2-5 garments changing in one video, use outfit_changes in appearance order after product_url. For one garment worn throughout with up to five separately sold outfit items, use styling_products with category tops, bottoms, outerwear, footwear, accessories or bags. These arrays are mutually exclusive. Both modes require a chosen model_id, reference_video_url, womens_apparel, 15 seconds, one English video. Use brief for order/occasion in outfit changes, or styling/placement direction in a styled outfit. Reference cut timing and visible accessory placement are checked before video generation; no equal-time slots or new actions are invented. Confirm the total quoted credits and pass expected_total_credits.

create_product_videos

Delete a post made from this account from LinkedIn. Cannot be undone.

delete_linkedin_post

Delete one of the account's own models and its photos, for good.

delete_custom_model

Disconnect this account's LinkedIn after the user confirms. Already published posts stay on LinkedIn.

disconnect_linkedin

Replace the text of a published LinkedIn post made from this account.

edit_linkedin_post

Whether the user's own LinkedIn is connected, as whom, days until it must be reconnected, and today's LinkedIn posts. Check before posting a video to LinkedIn.

get_linkedin