On faymas.in the catalogue is built from a very small vocabulary. Of 18,846 prompt URLs in the sitemap on 2026-10-17, the single word "portrait" appears in 7,097 of them (37.7%) and 7,783 (41.3%) mention a portrait, girl or woman. One title — "ultra realistic cinematic portrait of a young man" — is reused by 209 different pages, and those 209 pages carry 209 different prompts. Strip the 8-character ID from every slug and 3,892 URLs (20.7%) repeat a title already used elsewhere.
This guide analyses published prompt text. No generation tests were run.
What we counted
We took the sitemap, kept the 18,846 URLs under /prompt/, removed the trailing 8-hex ID from each slug, split the rest on hyphens, and counted how many slugs contain each word. Then we opened 40 of the pages that share a single title and read their prompt text out of the rendered HTML. No third party data was used; every number below is a count over that sitemap or those pages.
All 18,846 slugs end in an 8-character hex ID. Not one exception. That matters later.
The top of the vocabulary
| Word | Slugs containing it | Share of 18,846 |
|---|---|---|
| portrait | 7,097 | 37.7% |
| cinematic | 2,665 | 14.1% |
| prompt | 2,511 | 13.3% |
| young | 2,011 | 10.7% |
| realistic | 1,996 | 10.6% |
| ultra | 1,704 | 9.0% |
| man | 1,372 | 7.3% |
| couple | 1,269 | 6.7% |
| golden | 940 | 5.0% |
| aesthetic | 917 | 4.9% |
Group those words into families and the shape of the catalogue shows up:
| Family | Slugs with ≥1 member | Share |
|---|---|---|
| portrait / girl / woman | 7,783 | 41.3% |
| photo / cinematic / realistic / ultra | 4,594 | 24.4% |
| the word "prompt" itself | 2,529 | 13.4% |
| viral / instagram / reels / dp | 388 | 2.1% |
| Indian festivals (Diwali, Navratri, Dussehra…) | 318 | 1.7% |
| editing / retouch / face / background | 295 | 1.6% |
| resume / CV / passport / exam | 68 | 0.4% |
| named AI models (GPT, Midjourney, Nano Banana…) | 37 | 0.2% |
| products, brands, logos | 35 | 0.2% |
| the word "faymas" | 12 | 0.06% |
One title, 209 pages, 209 different prompts
This is the part we expected to be hollow and was not. 209 URLs share the body ultra-realistic-cinematic-portrait-of-a-young-man. We opened the first 40 and pulled the prompt out of the page text. 37 of the 40 gave us a clean read; the other 3 did not, and we are not counting them as missing.
Those 37 pages hold 37 distinct prompts. Median length 195 words, shortest 89, longest 476. Here are the first two, verbatim:
…"Create an ultra-realistic 4:5 cinematic mirror selfie of the same young Indian male with thick, medium-long messy layered shaggy hair…"
…"A young Indian man in his mid-20s, with stylish black hair, a light beard, wearing black sunglasses, light pastel yellow loose button-up shirt…"
Different pictures, different clothes, different settings. What is identical is the <title>: all 40 pages return "Ultra realistic cinematic portrait of a young man | Faymas".
So the duplication is not duplicated content. It is duplicated labeling. The person uploading wrote 209 different prompts and gave them all the same name.
That changes what the URL count means. 18,846 is a count of real, separate prompts — we found no empty or repeated prompt text in the sample — but the titles that describe them carry almost no information. A search engine reading those titles sees one sentence 209 times.
The other kind of repetition: names
Not every repeated slug body is a description. 509 of the 910 repeated bodies contain neither "prompt" nor "portrait" — they are person names and handles. kuldeep-meena covers 238 URLs on its own; vikash-chouhan another 70. Together, name-shaped repeats account for 1,892 URLs, 10.0% of the catalogue.
A slug that is just a creator's name tells a reader nothing about whether they want it. And a name collides with itself every time that creator uploads again, which is why one of them has 238.
Total repetition: 3,892 URLs (20.7%) sit under a title that at least one other URL already uses.
The brand is almost never in the URL
Twelve slugs of 18,846 contain "faymas" — 0.064%. Seven of them are literally faymas-xxxxxxxx with nothing else. This is what the person typed into a title field, not a naming convention. Two more are captions that got pasted into the title box: "follow me instagram and faymas please", "vipaarish100k my instagram id follow me prompt faymas on available ok".
And those twelve are not a random slice. They average 407 copies each, against a median of 34 across 50 randomly sampled prompt pages from the day before. Their best performer, one of the bare faymas-xxxxxxxx pages, has 1,566 copies.
We cannot say why. It could be that these early uploads sat on the homepage long enough to accumulate, or that the ones who typed the brand name into a title field were promoting themselves hardest. What we can say is that the only pages in the catalogue that carry the brand in the URL are also the pages with the highest copy counts we saw — and that the brand appears in 0.064% of the URLs a search engine can read.
What the tail cannot do
A catalogue this concentrated is not a broad library that happens to be big. 41.3% of it is portraits of people; 24.4% is the same five photographic adjectives; 2.1% is anything social-media flavored; 0.2% names a model. The instrument names — GPT, Gemini, Midjourney, Nano Banana — live on the page as a label, not in the URL, so 37 of 18,846 is the count of prompt-sellers who bothered to write the model name into their own title.
If you are looking for a faymas prompt for a product shot, a logo, a résumé or a passport picture, the sitemap offers 16, 14, 0 and 7 URLs respectively. The catalogue is large in one direction and thin in every other. That is a supply-side fact about the people uploading, not about what the site asks for — and it is the same shape as the interest profile: the copy counts cluster where the portraits cluster.
How this could be wrong
Slug tokens are not the prompt. Someone can write "saree" into a prompt without putting it in the title, and our family scores would miss it. We counted titles, not content, for the tables above — only the 40 pages we opened were read for their actual prompt text. And a repeated body is not proof the uploader was careless; a name is a reasonable title if the picture is of you.
What the numbers do establish is narrower and harder: the most common word in this catalogue's URLs is portrait, one sentence names 209 different pictures, and the brand shows up in 12 of 18,846.
FAQ
How many faymas prompts mention a portrait?
7,097 of 18,846 slugs contain the word portrait (37.7%), and 7,783 (41.3%) mention portrait, girl or woman.
Why do 209 faymas pages share the same title?
Because the title comes from what the uploader typed. We opened 40 of those 209 pages and found 37 distinct prompts — different pictures under one title.
How often does "faymas" appear in faymas.in URLs?
Twelve times out of 18,846 slugs, 0.064%. Seven of those are just faymas plus an ID.
How many faymas.in prompts name an AI model in the URL?
37 of 18,846 (0.2%). Model names appear on the page as a label but rarely in the slug.
Is the faymas catalogue repetitive content?
No. In the 40 pages we opened, every prompt text was different. What repeats is the title, not the content.