How to Use the Prompt Library and Prompt Builder: Copy a Ready-Made Prompt, Fill In the Blanks

Go to dijikocum.com/yapay-zeka/prompt. It's free, no sign-up.

The page has two doors: Image Prompts (10 prompts: photo sharpening, background replacement, combining photos, dressing a model, portraits, logos and posters) and Video Prompts (5 prompts: animating a photo, product video, short ad, venue tour). Tap the one for the job you want to do. The counts are as of November 1, 2026; the list is ordered "business-useful first, then trends."

The list page has three filter groups:

• What for: e.g. Background replacement, Combining photos, Campaigns and posters, Product photo (for video: Animating a photo, Product video, Promo and ads, ASMR and texture, Venue tour).
• Who it's for: Hotel and spa, Restaurant and café, E-commerce, Service business, Content creator.
• Aspect ratio: Square 1:1, Portrait 4:5, Photo 3:2 (videos are Full vertical 9:16).

The cards show whether it works with your own photo, the aspect ratio and how many attempts the example took. All the example images are the site's own output, and the card has a label saying so.

From top to bottom, the prompt page has:

• Before · input / After · result: the photo the prompt was given and the result that came out.
• Example prompt: the English text to give the model.
• Keep unchanged: the things the prompt asks the model not to change (e.g. the product's shape, count, color). According to the page's note, these are the things that break most often; check the result anyway.
• When not to use it: situations where the prompt won't help or where the result shouldn't be published.
• Production record: the model, production date, which attempt it came from, aspect ratio, quality and a What happened across attempts note.

For example, the "Change the background without altering the product" prompt was produced with GPT Image 2.5 on October 2, 2026, on the first attempt; the note says "The handles got slightly thinner and the color turned a bit grayer." In other words, the example isn't a guarantee; it shows what kind of result to expect.

Press the Copy prompt button; the text is copied to your clipboard. The gold [square-bracketed] parts are the ones you fill in: your product, your venue or your brand. Delete the square brackets and what's inside them, and write your own details instead. When describing the product, always write the count; the more detail you give, the less the model improvises.

Example: in the background prompt, you replace [the stack of four speckled ceramic mugs] with your own product (e.g. two glass jars of homemade fig jam with white lids) and [a sunlit kitchen windowsill ...] with the new scene. Here's the prompt on the site:

Go to dijikocum.com/yapay-zeka/prompt/olusturucu (every prompt page also has a Fill in the blanks tab, where you can do the same thing).

1. 01Pick an example: one of the 15 prompts, or "Image" or "Photo to video" under Or build from scratch.
2. 02Fill in the blanks, copy the prompt: write the square-bracketed parts in Turkish. You pick the words that change the result the most, like lighting, camera angle and movement, from a list; the English comes ready. Then copy the prompt to give the model.

What you need to know (from the page):

• Free text you write yourself isn't translated; it's shown in blue, and you get a translation prompt you can give to ChatGPT or Gemini.
• Text that will appear in the image and Turkish speech in the video are never translated; their letters are preserved.
• The builder runs entirely in your browser: what you type isn't sent to a server and isn't saved; it's deleted when you close the page.
• The site wrote object names once in Turkish and once in English and tried them on two models; both models drew the objects written in Turkish (like a copper cezve or a simit) correctly. That's a single trial per language and model; long scene descriptions and video models aren't covered by this trial.

Paste the prompt into an image-generating tool like ChatGPT or Gemini. If it's a prompt that works with your own product photo (the card says "With your own photo"), add your photo first, then send the prompt. You use video prompts in a video-generating tool.

When the result comes in, check:

• Count: is the number of products the same?
• Color and shape: have the product's color, handle, label or proportions been distorted?
• Text: are the Turkish characters on the poster (c-cedilla, g-breve, dotless i, s-cedilla, u-umlaut, o-umlaut) correct?

Even if you run the same prompt with the same settings, the result won't come out exactly the same. If the product itself has changed, don't use that image, as the site's "When not to use it" note says; generate it again. Before using it for sales, zoom in and compare it with the real product.

• Product photo: "Change the background without altering the product." Moves the product photo you have into a new scene; check the count and color.
• Café, restaurant: "Discount poster — tested for Turkish characters." A poster prompt tested for whether Turkish characters come out correctly; still read the letters one by one.
• Reels: "3-shot short ad video." A short ad with a timeline; you add sound and text afterward.

All of them are in the Prompt Library, for free.

• E-commerce seller: Prepare your main image with the "White-background product image for marketplaces" prompt and your social media image with "Change the background without altering the product." Be careful not to show the product's color or size as different from what it is.
• Restaurant: Use "Combine two photos: place the product in the venue" to put your dish into a photo of your own dining room. If the scene shows the food differently from what it is (ingredients that aren't in it, a size it doesn't really have), don't publish it.
• Boutique hotel or spa: Use "Hotel room promo video: slow push-in toward the window" to create a short video from a room photo; if something that isn't in the photo appears in the video, don't use it.
• Hairdresser or consultant: Use "Professional portrait from your own photo" to prepare a LinkedIn photo. The prompt is for your own photo; don't use it with someone else's face.

If you don't know which tool to run the prompt in, go to the Tool Atlas through the Image generation tools and Video generation tools links at the bottom of the prompt pages. If you'll use the image in an ad, read the link to the AI labeling rule on the prompt page.

Free, no sign-up. If you get stuck anywhere, message @dijikocum on Instagram.