AI image generation tools

{{whyLabel}}: Choosing the right tool prevents wasted effort and subscription costs.

{{howLabel}}:

• Use GPT-Image 1.5 (via ChatGPT) if you need the highest logical prompt adherence and conversational editing.
• Choose Midjourney V8 if your focus is on high-end artistic aesthetics and cinematic quality.
• Opt for Flux.2 (via Replicate or local) for the best photorealism and anatomical accuracy (hands/eyes).
• Use Adobe Firefly 5 for commercial projects requiring legal safety and Photoshop integration.

{{doneWhenLabel}}: You have a clear choice and an active account on one platform.

{{whyLabel}}: Proper setup ensures you aren't limited by interface constraints.

{{howLabel}}:

• For Midjourney: Join the Discord server or use the dedicated web alpha; use /settings to ensure 'V8' and 'High Variation Mode' are active.
• For GPT-Image: Ensure you are using the 'Plus' or 'Pro' model to access the DALL-E 3/GPT-Image 1.5 backend.
• For Flux/Stable Diffusion: Use a cloud provider like Fal.ai or Replicate for instant access without high-end hardware.

{{doneWhenLabel}}: You have successfully generated a test image using the command /imagine or a text box.

{{whyLabel}}: Vague prompts lead to generic results; structure provides control.

{{howLabel}}:

• Subject: Be specific (e.g., 'A weathered 70-year-old fisherman' instead of 'a man').
• Action/Setting: Describe the interaction (e.g., 'mending a neon-glowing net on a dark pier').
• Lighting/Camera: Add technical terms like 'Rembrandt lighting', '85mm lens', or 'low-angle shot'.
• Style: Reference specific aesthetics like 'Cyberpunk', 'Ukiyo-e', or 'Kodachrome 64'.

{{doneWhenLabel}}: You have generated 5 variations of a single concept by adjusting only one variable at a time.

{{whyLabel}}: Parameters allow you to control technical aspects like aspect ratio and unwanted elements.

{{howLabel}}:

• Use --ar 16:9 or --ar 9:16 to set the aspect ratio (crucial for social media vs. cinematic shots).
• Use --no (in Midjourney) or the 'Negative Prompt' field (in Flux/SD) to exclude elements like 'blurry, text, low quality'.
• Apply --stylize (values 0-1000) to control how much the AI applies its own artistic 'opinion'.

{{doneWhenLabel}}: You have produced a widescreen (16:9) image that successfully excludes a specific color or object.

{{whyLabel}}: Maintaining the same character across different scenes is the 'Holy Grail' of AI storytelling.

{{howLabel}}:

• In Midjourney: Use the --cref [URL] parameter followed by the link to your 'base' character image.
• Adjust --cw (Character Weight) from 0 to 100; 100 keeps the face and clothing, 0 focuses only on the face.
• In Flux/SD: Use a 'LoRA' (Low-Rank Adaptation) specifically trained on a character if using local tools.

{{doneWhenLabel}}: You have 3 images of the same character in 3 different environments (e.g., a forest, a city, a space station).

{{whyLabel}}: AI often makes small mistakes (e.g., extra fingers); inpainting allows you to fix only the error.

{{howLabel}}:

• Select the 'Vary Region' or 'Inpaint' tool in your chosen software.
• Mask (paint over) only the area you want to change (e.g., a hand or a background object).
• Provide a new, simple prompt for that specific area (e.g., 'hand holding a coffee cup').

{{doneWhenLabel}}: You have successfully modified a small part of an existing image without changing the rest of the composition.

{{whyLabel}}: Local tools like Flux.2 or Stable Diffusion offer the most control but are usually hard to install.

{{howLabel}}:

• Download Pinokio (open-source browser for AI).
• Search for 'Flux.1-dev' or 'Forge UI' within the Pinokio browser.
• Click 'Download' and let it handle the Python/Git dependencies automatically.
• Ensure you have at least 12GB of VRAM (NVIDIA RTX 3060 or better) for smooth performance.

{{doneWhenLabel}}: The local WebUI opens in your browser and generates an image without an internet connection.

{{whyLabel}}: Moving from single images to a series proves mastery of consistency and narrative.

{{howLabel}}:

• Define a short story (e.g., 'A robot discovering a flower in a wasteland').
• Use your Character Reference to keep the robot consistent.
• Use a Style Reference (--sref in Midjourney) to ensure the lighting and color palette match across all 5 pages.
• Upscale the final images to 4K using an AI upscaler (like Magnific or the built-in Midjourney upscaler).

{{doneWhenLabel}}: You have a PDF or gallery of 5 high-resolution, stylistically consistent images telling a story.