Title:
Optimizing AI Image Generation on Low-Resource Devices (₹6400 Mobile) – Prompt Engineering Insights
Body:
Hello Developers,
I would like to share an experimental case study and seek insights from the community.
Over the past few weeks, I have been working extensively on AI image generation using a very low-resource setup — specifically a ₹6400 4G smartphone (itel A50). Despite the hardware limitations, I was able to generate a large number of high-quality images by focusing heavily on prompt engineering rather than computational power.
Key Observations:
Prompt Structure Matters More Than Hardware
Carefully layered prompts (subject + environment + lighting + cinematic style) significantly improved output quality.
Iterative Prompt Refinement
Instead of regenerating randomly, making small, controlled modifications to prompts yielded more consistent results.
Token Efficiency
Short but semantically dense prompts performed better than long descriptive ones in some cases.
Style Anchoring
Reusing specific stylistic keywords helped maintain visual consistency across multiple generations.
Challenges Faced:
Inconsistent output quality across similar prompts
Difficulty in maintaining character continuity
Limited control over fine details (hands, proportions, background coherence)
Questions for the Community:
What are the best practices to improve output consistency across multiple generations?
Are there any prompt structuring frameworks or patterns that work reliably across different scenarios?
How can we better control character continuity and fine details using prompt engineering alone?
Are there any recommended techniques for optimizing results when working from low-resource devices?
I believe this area has strong potential, especially for creators working without access to high-end systems.
Looking forward to your insights and suggestions.
Thank you.
Eswar. G. Guni mahamuni