Troubleshooting generations

Even with well-written prompts, AI generation can sometimes produce unexpected results. This guide helps you identify and fix common issues.

Common problems and solutions

Blurry or low-quality output

Symptoms: Soft focus, lack of detail, muddy textures

Causes:

• Prompt lacks quality keywords
• Conflicting style requests
• Overly complex scene

Solutions:

• Add quality keywords: 4K, high resolution, sharp focus, detailed
• Specify professional photography or studio quality
• Simplify the scene - remove unnecessary elements
• Use crisp or clear in your prompt

Example fix:

• Before: "Woman in room"
• After: "Woman in room, professional photography, sharp focus, 4K, detailed"

Strange artifacts or distortions

Symptoms: Odd shapes, extra limbs, warped features, impossible anatomy

Causes:

• Contradictory instructions
• Overly complex poses
• Multiple subjects creating confusion

Solutions:

• Simplify the pose or action
• Focus on one subject at a time
• Be more specific about positioning
• Avoid complex hand positions (common AI weakness)
• Use anatomically correct if needed

Example fix:

• Before: "Woman doing yoga headstand while juggling"
• After: "Woman in simple yoga pose, standing, arms at sides"

Wrong composition or framing

Symptoms: Subject cut off, wrong angle, too much empty space

Causes:

• Lack of framing keywords
• Ambiguous positioning

Solutions:

• Add framing keywords: close-up, portrait, full body, waist up
• Specify camera angle: from above, eye level, from below
• Use centered or rule of thirds composition
• Add well-framed or professional composition

Example fix:

• Before: "Woman at beach"
• After: "Woman at beach, full body shot, centered composition, eye level"

Lighting issues

Symptoms: Too dark, too bright, flat lighting, harsh shadows

Causes:

• No lighting specified
• Contradictory lighting terms
• Unrealistic lighting setup

Solutions:

• Be specific: soft diffused lighting, studio lighting, natural window light
• Avoid contradictions: don't mix "bright" and "dark"
• Reference known setups: three-point lighting, Rembrandt lighting
• Add direction: from left, backlit, front-lit

Example fix:

• Before: "Woman, dramatic scene"
• After: "Woman, dramatic side lighting, film noir style, high contrast"

Unwanted elements or clutter

Symptoms: Extra objects, messy background, unintended additions

Causes:

• Vague environment description
• AI filling gaps with assumptions

Solutions:

• Specify minimal background or simple background
• Describe the exact setting: plain white background, clean bedroom
• Use isolated or focused on subject
• Add uncluttered or minimal

Example fix:

• Before: "Woman indoors"
• After: "Woman indoors, minimal background, clean modern bedroom, uncluttered"

Wrong mood or energy

Symptoms: Result doesn't match intended feeling

Causes:

• Conflicting mood keywords
• Style keywords overwhelming mood intent

Solutions:

• Be explicit about mood: playful, sensual, intense
• Add expression keywords: smiling, sultry expression, intense gaze
• Use [mood] atmosphere or [mood] energy
• Check that lighting supports the mood

Example fix:

• Before: "Woman portrait"
• After: "Woman portrait, sensual expression, intimate atmosphere, soft romantic lighting"

Systematic troubleshooting

When results aren't working:

Step 1: Identify the issue

• Is it technical (blur, artifacts)?
• Is it compositional (framing, angle)?
• Is it aesthetic (style, mood)?
• Is it content (wrong subject/setting)?

Step 2: Isolate the cause

• Remove all style keywords and try basic prompt
• Add keywords back one at a time
• Note which addition causes issues

Step 3: Fix and test

• Apply targeted solution from above
• Generate and compare
• Iterate if needed

Step 4: Build on success

• Once it works, save that prompt pattern
• Use as template for similar content
• Document what keywords worked

Video-specific issues

Videos have unique challenges:

Motion problems

• Issue: Jerky movement, unnatural motion
• Fix: Use smooth motion, slow movement, gentle

Consistency issues

• Issue: Subject changes mid-clip
• Fix: Simpler prompts, fewer style keywords, focus on subject details

Quality degradation

• Issue: Video looks worse than image previews
• Fix: Add high quality video, stable, consistent

Prevention tips

Avoid issues before they happen:

1. Start simple: Basic prompt first, add complexity gradually
2. Reference test: Try similar prompt variations
3. Check spelling: Typos can cause unexpected results
4. Read examples: Study working prompts from others
5. One change at a time: When iterating, modify one element only

When to contact support

Some issues may be platform-related:

• Consistent failures across different prompts
• Technical errors or credit deduction without output
• Severe quality degradation from previous sessions
• Content policy violations being incorrectly flagged

Contact support@goon.com with:

• Your prompt text
• Description of the issue
• Screenshot if applicable

FAQs

Why do hands often look wrong?

Hands are notoriously difficult for AI. Minimize hand visibility or keep them in simple positions (at sides, on hips, behind back).

Can I regenerate without using more credits?

No, each generation attempt uses credits. Write careful prompts to minimize wasted generations.

Why do results vary even with the same prompt?

AI generation includes randomness. Each attempt produces different results even with identical prompts.

Related guides

• Prompt writing basics
• Style keywords and examples
• Iterative workflow and variations