Hi everyone,
I built ImgExtender because I kept running into the same daily frustrations when working with images — AI-generated pictures not fitting the correct aspect ratio, blurry screenshots, compressed images losing too much detail, background edits taking too long, and having to jump across 4–5 different tools for what should have been simple tasks.
So I decided to create a single lightweight web tool that solves the most common image problems without requiring Photoshop skills, plugins, or GPU setup.
https://imgextender.com/
Below is a quick introduction to the problem, what ImgExtender does, and what’s behind it.
Why I Built It
Most people don’t need a full graphics suite. They just want to quickly:
Extend an image (AI outpainting) without weird borders
Upscale or unblur without making the picture look “AI smooth”
Remove background cleanly (not the soft-edge mess from traditional tools)
Fix awkward aspect ratios for social media
Compress large images without ruining quality
Do simple edits (crop/resize/filters/text) without learning complex software
But the existing tools are either:
too slow,
too complicated,
too expensive,
or too fragmented (one tool for background removal, another for upscaling, another for AI generation…).
I wanted one place that makes image tasks fast, predictable, and accessible.
What ImgExtender Does (Core Features)
All tools work in-browser, no signup required.
AI Image Tools
Extend Image (Outpainting) — expand canvas naturally using generative fill
AI Generate Image — text-to-image with clean, sharp output
Unblur Image — restore clarity using deep-learning deblurring
Upscale Image — 2×–8× enlargement with texture recovery
Editing Tools
Crop
Resize
Rotate & Flip
Add Text
Filters
Blur Background
Remove Background
Change Background
Utility Tools
Image Compressor
Image to PDF converter
It’s designed to be something even a non-technical person can use instantly.
Under the Hood (Technical Notes)
Outpainting uses a fine-tuned diffusion model that blends edge areas to avoid seam artifacts.
Unblur uses a multi-stage sharpness restoration pipeline (deconvolution + learned edge enhancement).
Upscaling uses a lightweight ESRGAN variant optimized for smoother detail recovery.
Background removal uses a segmentation model focusing heavily on hair & soft edges.
All processing is GPU-accelerated on the backend; front-end is minimal and optimized for speed.
I’m also working on client-side WASM acceleration for some tasks.
Who Uses It Today?
From early usage logs and feedback:
AI artists fixing midjourney/dalle artifacts
Designers preparing images for clients
Etsy sellers enhancing product photos
Developers who need quick assets without opening Photoshop
Students / teachers improving slides or PDFs
Casual users cleaning up old photos
Roadmap
Batch processing
API access
A unified “smart enhance” mode
More style-aware AI generation
Client-side processing to reduce server load
If you use AI image tools often, I’d love to hear what your workflow looks like.
Feedback Wanted
This is still early-stage.
I would really appreciate feedback from the HN community on:
What feels slow or confusing
Which features matter the most
What is missing for real professional use
Whether you'd want self-hosting or local models
Your typical image-edit workflow
Link: https://imgextender.com/
No login required.
Thanks for reading — and thanks in advance for any feedback!
Why I Built It Most people don’t need a full graphics suite. They just want to quickly:
Extend an image (AI outpainting) without weird borders
Upscale or unblur without making the picture look “AI smooth”
Remove background cleanly (not the soft-edge mess from traditional tools)
Fix awkward aspect ratios for social media
Compress large images without ruining quality
Do simple edits (crop/resize/filters/text) without learning complex software
But the existing tools are either:
too slow,
too complicated,
too expensive,
or too fragmented (one tool for background removal, another for upscaling, another for AI generation…).
I wanted one place that makes image tasks fast, predictable, and accessible.
What ImgExtender Does (Core Features) All tools work in-browser, no signup required. AI Image Tools
Extend Image (Outpainting) — expand canvas naturally using generative fill
AI Generate Image — text-to-image with clean, sharp output
Unblur Image — restore clarity using deep-learning deblurring
Upscale Image — 2×–8× enlargement with texture recovery
Editing Tools
Crop
Resize
Rotate & Flip
Add Text
Filters
Blur Background
Remove Background
Change Background
Utility Tools
Image Compressor
Image to PDF converter
It’s designed to be something even a non-technical person can use instantly.
Under the Hood (Technical Notes)
Outpainting uses a fine-tuned diffusion model that blends edge areas to avoid seam artifacts.
Unblur uses a multi-stage sharpness restoration pipeline (deconvolution + learned edge enhancement).
Upscaling uses a lightweight ESRGAN variant optimized for smoother detail recovery.
Background removal uses a segmentation model focusing heavily on hair & soft edges.
All processing is GPU-accelerated on the backend; front-end is minimal and optimized for speed.
I’m also working on client-side WASM acceleration for some tasks.
Who Uses It Today? From early usage logs and feedback:
AI artists fixing midjourney/dalle artifacts
Designers preparing images for clients
Etsy sellers enhancing product photos
Developers who need quick assets without opening Photoshop
Students / teachers improving slides or PDFs
Casual users cleaning up old photos
Roadmap
Batch processing
API access
A unified “smart enhance” mode
More style-aware AI generation
Client-side processing to reduce server load
If you use AI image tools often, I’d love to hear what your workflow looks like.
Feedback Wanted This is still early-stage. I would really appreciate feedback from the HN community on:
What feels slow or confusing
Which features matter the most
What is missing for real professional use
Whether you'd want self-hosting or local models
Your typical image-edit workflow
Link: https://imgextender.com/ No login required. Thanks for reading — and thanks in advance for any feedback!