Back to WorkspaceDeep Stencil
Deep Stencil Logo

User Guide

Everything you need to know to create stunning tattoos with Deep Stencil.

Getting Started

Deep Stencil offers three creation modes, each designed for a different creative approach. Choose the one that best fits your idea.

Text

Describe your tattoo in words and the AI generates it for you.

Photo

Upload reference images to guide the generation.

Edit

Refine an existing design with text instructions.

Text Mode

Text Mode is the most versatile: describe your ideal tattoo and the AI creates it. You can combine the prompt with preset styles or style reference images to guide the aesthetic of the result.

The Text Prompt

The prompt is the heart of Text Mode: describe subject, composition, mood, and what you DON'T want. The more details you provide, the better the result. Full example: 'Roaring realistic lion with detailed mane, vertical composition, smoke background, no colors, black and white only'.

How to write a good prompt

  • Be specific: 'Japanese dragon with waves and cherry blossoms' works better than 'dragon'.
  • Indicate the composition: 'vertical', 'circular', 'wrapped around the arm'.
  • Specify what you DON'T want: 'no text', 'no colors', 'no background'.
  • Add mood details: 'dark style', 'elegant and minimalist', 'aggressive'.

How to Choose the Style

You have two options to define the style, and they are mutually exclusive:

Preset Style — Select one of 10 styles from the grid. Parameters (rendering, detail, fidelity) are automatically optimized for that style.

Reference Image — Upload 1-2 photos of tattoos or drawings with the style you want. The AI will analyze the aesthetic and apply it to your generation.

Selecting a preset style automatically removes reference images, and vice versa. This ensures the AI receives a clear and unambiguous stylistic direction.

Important: style reference images must be tattoos or drawings/illustrations, NOT regular photos (landscapes, selfies, objects). The AI needs to see a concrete artistic style to analyze and replicate. A photo of a realistic tattoo, a digital illustration, or a sketch work perfectly.

Warning: if you use a COLOR tattoo photo as a style reference, you must select "With colors" in the RENDERING section of the parameters. Otherwise the AI will generate the design in black and white, ignoring the colors from the reference.

Prompt Fidelity

The slider controls how literally the AI follows your text description:

0-30% — The AI follows the prompt literally, producing exactly what you described. Use this range when you have a precise idea.

40-60% — Good balance between fidelity and creativity. Recommended for most uses.

70-100% — The AI reinterprets the prompt with artistic freedom. Perfect for discovering unexpected variations and original ideas.

Ideal for: those who have a clear idea to describe in words, those who want to explore different styles quickly, those starting from scratch without reference images.

Photo Mode

Photo Mode lets you guide the AI with reference images instead of a text prompt. Each image has a specific role that determines how it will be used.

How It Works

Unlike Text Mode, you don't write a prompt here. Images guide the AI. Upload photos and assign each one a role by clicking on the colored badge. The AI will interpret each image based on its assigned role.

max 5
Subject

The main image: the tattoo subject. Upload up to 4 photos of the desired subject.

Style in Photo Mode

The tattoo style is managed from the STYLE section below, identical to Text Mode. You can choose a preset style or upload a style reference image. Subject photos define WHAT to generate; the style defines HOW to generate it.

Warning: if you use a COLOR tattoo photo as a style reference, you must select "With colors" in the RENDERING section of the parameters. Otherwise the AI will generate the design in black and white, ignoring the colors from the reference.

Image Fidelity

The slider controls how closely the AI follows the uploaded photos:

0-30% — The AI reproduces the photos as faithfully as possible, maintaining shapes, proportions, and details. Ideal for turning a real photo into a tattoo.

40-60% — The AI uses the photos as a primary base but takes artistic liberties. The most balanced range.

70-100% — The AI uses the photos only as initial inspiration, reinterpreting them creatively. Perfect for unique and surprising results.

Tips for better photos

  • Use sharp, well-lit images — avoid blurry or dark photos.
  • For the subject, isolate it as much as possible (simple background).
  • For style, choose existing tattoos that represent the desired aesthetic.

Ideal for: turning real photos into tattoos, combining multiple subjects into one design, those who prefer visual communication.

Edit Mode

Edit Mode lets you take an already generated design (from the gallery or uploaded) and modify it with text instructions. Perfect for touch-ups and variations.

Image to Edit

Upload the image you want to modify. You can select a design from the gallery (original parameters are automatically restored) or upload a file from your device.

Reference Image (Optional)

You can upload an additional image to show the AI the direction of the edit. Example: upload a blackwork tattoo to show the AI you want to convert your design to that style.

The Edit Prompt

Describe precisely what you want to change. Effective commands: 'add deeper shadows', 'remove the background', 'convert to black and white', 'add flowers around the subject'. Avoid vague descriptions like 'make it better'.

How to get precise edits

  • Be specific about what to change: 'add more detail to the dragon scales'.
  • You can also upload a reference image to guide the edit.
  • For major changes, it's better to regenerate from scratch rather than editing.

Ideal for: retouching already generated designs, adding or removing specific elements, changing rendering style, perfecting details of an almost-perfect design.

Available Styles

Deep Stencil offers 10 preset styles, each with automatically optimized parameters. You can also upload reference photos for a custom style.

Traditional

Thick lines, bold colors, classic subjects (roses, eagles, anchors). The most iconic Western tattoo style.

Realistic

Detailed shadows and realistic gradients. Ideal for portraits, animals, and photorealistic still life.

Japanese

Waves, dragons, koi fish, cherry blossoms. Wide, flowing compositions typical of Japanese irezumi.

Neo-Traditional

Evolution of Traditional with more details, rich gradients, and extended color palette.

Blackwork

Black only. Geometries, tribal patterns, mandalas. Perfect for high-contrast, high-impact designs.

Fine Line

Ultra-thin, delicate lines. Ideal for minimalist tattoos, lettering, stylized flowers, and discreet designs.

Cyber Sigilism

Flowing, abstract lines inspired by sigils and circuits. Futuristic and mystical aesthetic.

Micro-Realism Vertical

Micro-realism in vertical format. Perfect for forearm, ribcage, and elongated compositions.

Micro-Realism

Extreme realism at a reduced scale. Incredible details concentrated in a small space.

Conceptual

Artistic and experimental style. Abstract and surreal interpretations of the subject.

When you select a style, parameters (rendering, detail, format) are automatically optimized. You can always adjust them manually afterward.

When to Use Preset Styles

Use a preset style when you want a quick, optimized result. Each style automatically configures rendering (outline/shaded/color), detail level, and fidelity for optimal results with that aesthetic. Click a style to select it, click again to deselect.

Preset Style vs Reference Image

If you have a precise aesthetic in mind that doesn't match the 10 presets, upload a style reference image. The AI will analyze the artistic style from the photo and apply it to the generation. The two options are exclusive: selecting a preset removes reference images, and vice versa.

Remember: the style reference photo must be a tattoo or a drawing/illustration, not a regular photo. The AI needs to analyze a concrete artistic style.

Parameters

Each generation can be customized with these parameters. The selected style sets optimal values, but you can adjust them as you like.

Rendering Mode

Outline (lines only), Shaded (lines + shadows), Color (with colors). For stencils, use Outline. For realistic tattoos, use Shaded.

Detail Level

Low (fast, essential lines), Medium (balanced), High (maximum detail). Use High for complex designs and Low for quick sketches.

Aspect Ratio

Square (1:1) for symmetric designs or free placement. Vertical (9:16) for elongated designs like forearm or leg.

Prompt / Image Fidelity

The slider controls the balance between input fidelity and AI creative freedom. In Text Mode it's called 'Prompt Fidelity', in Photo Mode 'Image Fidelity'.

This is the most important parameter for controlling the result:

  • In Text Mode: low values = the AI follows the prompt literally. High values = the AI reinterprets creatively.
  • In Photo Mode: low values = the AI reproduces photos faithfully. High values = the AI reinterprets photos freely.

0-30%: Maximum fidelity. Predictable result that closely matches the input. Ideal when you know exactly what you want.

40-60%: Optimal balance between fidelity and creativity. Recommended starting point.

70-100%: Maximum creative freedom. Original but less predictable results. Perfect for exploring ideas.

Tip: start with the default value and adjust after the first result. If it's too generic, lower it. If it's too literal, raise it.

Body Part

Select where you want to place the tattoo. This helps the AI generate a design with proportions and composition suited to that area.

Try On Body

See how the tattoo will look on your body before going to the studio. The AI blends the design onto the photo realistically.

  1. Generate a design (or select one from the gallery).
  2. Open the 'Try On Body' panel on the right.
  3. Upload your own photo or choose a body template.
  4. Position, resize and rotate the tattoo on the photo.
  5. Click 'Generate Try-On' — the AI will blend the design realistically.

For a better Try-On

  • Use photos with clearly visible skin and even lighting.
  • Avoid photos with clothes covering the tattoo area.
  • Manual positioning is precise — move and resize carefully before generating.

Possible blending artifacts

The AI that blends the tattoo onto the skin may occasionally produce visual artifacts or unexpected results (e.g., distortions, altered colors, or elements that weren't there). This is normal behavior for image generation models. If this happens, simply try again: click 'Generate Try-On' again, or slightly move or resize the tattoo before regenerating. A few attempts are usually enough to get a realistic result.

Controls

Desktop (mouse + keyboard)
  • Dragmove the tattoo
  • Ctrl+Scrollresize
  • Alt+Scrollrotate
Mobile / Tablet (touch)
  • 1 fingermove the tattoo
  • Pinchresize
  • 2 fingersorient

3D Preview

3D Preview lets you visualize the design on an interactive three-dimensional model, without using a real photo.

  • Rotate the model by dragging with the mouse.
  • Zoom with the mouse wheel.
  • Choose male or female body type.
  • Use the vertical slider to move the focus point.

3D Preview is completely free and doesn't consume any generations.

Controls

Desktop (mouse + keyboard)
  • Ctrl+Clickplace the tattoo on the model
  • Dragrotate the model
  • Scrollzoom in/out
  • Alt+Scrollrotate the tattoo on the model
Mobile / Tablet (touch)
  • Long pressplace the tattoo on the model
  • 1 fingerrotate the model
  • Pinchzoom in/out

Export Stencil

Stencil export transforms your design into a template ready for skin transfer. Available for Tattooist plans.

Canny (Edges)

Extracts design edges with precision. Ideal for tattoos with sharp, defined lines.

Halftone (Dot Pattern)

Converts gradients into printable dots. Perfect for designs with shadows and gradients.

Morphology (Simplified)

Reduces the design to essential contours. Useful for quick transfers and minimalist designs.

You can export as PNG or PDF, adjust DPI (300 or 600), contrast, brightness, and enable mirror for transfer.

AI models are stochastic

AI generation models produce probabilistic results: each generation is unique and may differ from previous ones even with the same settings and prompt. The AI may sometimes not precisely follow the text instructions or selected settings. If the result is not what you expected, simply try again — or slightly adjust the Prompt/Image Fidelity slider. A few more attempts are usually enough to get the design you want.

AI Limitations

All AI image generation models (including the one used by Deep Stencil) are subject to automatic safety filters enforced by providers. These filters may block or slow down generation in certain cases.

What can be blocked

  • Realistic human faces — AI models have restrictions on generating recognizable or photorealistic faces.
  • Human bodies — scenes showing nudity, suggestive poses, or anatomical content may trigger the filters.
  • Copyrighted characters — characters from movies, anime, video games, comics, or registered trademarks may be rejected by the model.

What happens in practice

The generation may fail completely and return an error.

The generation may take longer than usual (several extra seconds) while the model attempts to produce a compliant result.

The result may differ significantly from what was requested, as the model automatically modifies problematic elements.

Tips

  • If generation fails, try rephrasing the prompt avoiding direct references to faces, nudity, or copyrighted characters.
  • Use generic descriptions instead of specific names (e.g., 'armored warrior' instead of a character's name).
  • If the result differs from what you expected, try again — each generation is unique and a new attempt may produce better results.

These limitations are NOT caused by Deep Stencil but are inherent to all AI image generation models currently available on the market.

Pro Tips

Follow these tips to get the best possible results from Deep Stencil.

Prompt Writing

  • More details is better: size, position, mood, specific elements.
  • Combine style + subject: 'realistic lion in Japanese style with waves'.
  • If the result doesn't convince you, try rephrasing the prompt or changing the style.

Using Photos

  • Reference photos work best when they are sharp and well-lit.
  • If you upload a style reference, the AI will prioritize it over the preset style.
  • For optimal results in Photo Mode, use at least 1 subject photo + 1 style photo.

Try-On and Preview

  • Try different positions before deciding — placement greatly affects the final result.
  • 3D Preview is perfect for an overview, Try-On for realism.
  • Save Try-Ons you like to the gallery to compare them later.
  • If blending produces artifacts or strange results, try again or slightly move the tattoo — the AI produces different results with each attempt.