Logo for AiToolGo

Midjourney Advanced Guide: Generate Your Ideal AI Character Images in 7 Key Dimensions

In-depth discussion
Easy to understand
 0
 0
 5
This article, by the designer 解决师Blue, presents an advanced Midjourney tutorial for generating AI character images aligned with specific intentions. Drawing on photography knowledge consulted via ChatGPT and a photography-savvy friend, it systematically covers seven dimensions: camera viewpoint (front/back/side/low/high/bird's-eye), character angle, shot distance (ECU to full body), facial expressions, lighting for atmosphere (warm, soft, rim, neon, Rembrandt), clothing design via a fixed prompt formula, and art styles (realistic, acrylic, 3D blind-box, ink wash, Chinese illustration, pixel art). The author shares concrete prompt examples, observed results, and a closing formula combining subject description, viewpoint, angle, distance, expression, clothing, style, and light source. Key limitations and tips about prompt length and stepwise generation are also noted.
  • main points
  • unique insights
  • practical applications
  • key topics
  • key insights
  • learning outcomes

  • • main points

    • 1
      Organizes AI character generation into a clear, seven-dimension framework that maps to real photography concepts
    • 2
      Provides a reusable composite prompt formula that combines subject, viewpoint, angle, distance, expression, clothing, style, and light
    • 3
      Shows side-by-side comparisons of keyword variations (e.g., low angle view vs. plain low-angle keyword, Smiling vs. happy) that reveal how wording changes output

  • • unique insights

    • 1
      Distinguishes between 'viewpoint' (camera position relative to subject) and 'angle' (camera height), and shows that combining them (e.g., high angle + bird's-eye) yields more natural results than either alone
    • 2
      Notes that Midjourney interprets 'low/high angle' as depicted subject perspective rather than photographic camera angle, so an additional angle keyword is needed to disambiguate

  • • practical applications

    • Gives Midjourney users a structured, repeatable method to control composition, expression, lighting, clothing and style when generating character imagery, particularly useful for banner and marketing design work where consistent framing is needed. Readers can directly adopt the keyword libraries and the composite prompt template.

  • • key topics

    • 1
      Camera viewpoint and angle control (front, back, side, low, high, bird's-eye) for character composition
    • 2
      Shot distance terminology (extreme closeup to full length shot) and facial expression keywords
    • 3
      Lighting keywords (warm, soft, rim, neon, Rembrandt) for mood and a composite prompt formula combining style, clothing and art direction

  • • key insights

    • 1
      Translates photography vocabulary into concrete Midjourney prompt keywords, bridging a common knowledge gap for AI image creators
    • 2
      Offers a complete, reusable prompt template plus a library of tested keywords for viewpoint, distance, expression, lighting, clothing and styles
    • 3
      Documents empirical test results, including failures and workarounds, giving readers realistic expectations

  • • learning outcomes

    • 1
      Control camera viewpoint, angle and shot distance in Midjourney character images using photography-based keywords
    • 2
      Apply lighting keywords with weights and a composite prompt formula to build consistent, art-directed character prompts
    • 3
      Select and combine art styles and clothing descriptors to produce stylized character imagery suited to banner and marketing work
examples
tutorials
code samples
visuals
fundamentals
advanced content
practical tips
best practices

“ Introduction: Why AI Character Generation Needs Control

When beginners first use Midjourney to generate people, the biggest problem is that the position, scale, and viewpoint of the character cannot be guaranteed. Without constraints, the default output often leans toward a Western look with inconsistent framing. The solution is to borrow knowledge from photography and translate it into prompt keywords. This guide, based on hands-on testing, breaks character generation into seven controllable dimensions so that results become predictable and closer to the creator's intent.

“ Viewpoint Selection: Front, Back, Side, Low, High and Bird's-Eye Views

Viewpoint describes where the camera sits relative to the subject. A baseline prompt such as 'young girl' produces mixed angles. Adding 'front view' yields mostly frontal portraits, 'back view angle' produces figures seen from behind with a sense of mystery, and 'side view' delivers profiles that highlight contour and character. Low-angle and bird's-eye prompts can produce distorted or unnatural results when used alone, because the model interprets them as the figure's own vertical posture rather than the camera's position. This is why viewpoint must be combined with camera angle and distance. Other options such as free camera, fixed camera, dynamic, and perspective can be explored as well.

“ Camera Angle Selection: Low, High and Eye-Level Angles

Camera angle is the height of the lens. A low angle makes the subject look taller and more imposing, suited to architecture or heroic portraits. A high angle places the camera above and emphasizes the subject's size relationship with the ground, suited to crowds or cityscapes. An eye-level angle sits at the subject's height and feels neutral, ideal for portraits and landscapes. The author found that pairing 'low angle view' with the subject prompt resolved the unnatural 'looking at the camera' feel seen with viewpoint keywords alone, and that combining high angle with bird's-eye view looked more natural than bird's-eye alone.

“ Shot Distance Selection: From Extreme Close-Up to Full Body

Shot distance determines how much of the subject fills the frame. Extreme close-up (ECU) focuses on the face for detail and emotion. 'Headshot' frames from the top of the head to the neck. 'Chest shot' and 'waist shot' (also bust portrait) show upper body lines and expression. 'Medium full shot' cuts around the knees and balances the figure with its surroundings. 'Full length shot' shows the entire body, highlighting silhouette and outfit. Choosing the right distance is essential because it defines what the viewer focuses on, whether the face, the posture, or the clothing.

“ Facial Expression Selection: Controlling Emotion in Generated Portraits

Expressions convey emotion and mood. Keywords such as 'smiling' produce subtle facial changes, while 'happy' can also introduce body language changes, showing that different words yield different results. If the outcome is not ideal, switching to a synonym is a practical strategy. 'Angry' works well for intense expressions. 'Crying' is harder: tears appear only occasionally, so it usually needs additional descriptive keywords. Other useful expressions include afraid, jealous, confused, confident, disgust, and surprise, which can be explored freely.

“ Atmosphere and Lighting: Using Light to Shape Mood

Lighting strongly affects dimension and emotional tone. Many beginners ignore lighting, but specifying it adds depth. 'Warm light' creates a golden, gentle look, and its weight should be adjusted with the :: syntax, since stronger weight increases the sense of light source. 'Soft light' produces even, diffuse illumination, and a moderate weight usually looks better than the maximum. 'Rim light' creates bright edges that separate the subject from the background, common in film posters. 'Neon light' adds colorful, dramatic reflections. 'Rembrandt light' produces a classic one-side illumination with shadow on the other side, adding a dramatic, three-dimensional feel. The author notes that these are reference points for AI prompting rather than professional photography guidance.

“ Clothing Design: A Universal Prompt Formula for Outfits

Clothing details require knowledge of fashion, but a fixed structure helps: Theme + Style + Material + Element + Decorative details. Theme includes hanfu, long down coat, dress, or gown. Style includes minimalist, modern, retro, punk, British subculture, Renaissance, or Rococo. Material includes cotton, wool, silk, leather, or craftsmanship. Elements include shape, texture, color, pattern, and accessories. In the test, 'Chinese dress' alone produced plain, colorless outfits, while adding fabric, embroidery, diamond-encrusted craftsmanship, and floral motifs produced richly detailed results.

“ Art Style Exploration: Realistic, Acrylic, 3D, Ink, Chinese and Pixel Styles

Styles can be simple or highly layered. Realistic portraits such as a 1990s Hong Kong street couple need no special style keyword; detailed descriptions of the people, scene, and camera parameters are enough. 'Acrylic painting' blends colorful, futuristic, and 2D/3D elements. 'Popmart blind box' gives a clay-like 3D toy feel. 'Ink painting' combined with an artist reference and bold strokes creates a striking concept look. 'Extremely bold outline' with Chinese elements produces Chinese-style illustration; the author stresses that Chinese motifs and ink or watercolor cues must be emphasized, otherwise the result drifts into anime. 'Pixel art' prompts are limited, since Midjourney cannot produce true vector pixel art and instead creates pixelated imagery.

“ Conclusion: The Universal Prompt Formula and Iterative Workflow

A satisfying character image depends on precise wording covering the subject description, viewpoint, angle, distance, expression, clothing, style, and light source. Partial combinations can still yield good results. Crucially, Midjourney rarely achieves the desired effect in one prompt. Stuffing many keywords into a single prompt often hurts results, so the recommended approach is to build the image step by step, refine through variations, and use local redrawing for details such as facial features. Clear, selective keywords outperform long, cluttered ones.

 Original link: https://www.uisdc.com/character-synthesis

Comment(0)

user's avatar

      Similar Learning

      Related Tools