y-vladimir / smartdeblur

Restoration of defocused and blurred photos/images

SmartDeblur 1.27

SmartDeblur is a tool for restoring defocused and blurred images.
Written in C++ using Qt 4.8.
Algorithm based on several deconvolution techniques (Wiener, Tikhonov, Total Variation prior). Supported defect types are:

• Out of Focus blur (with kernel deep tuning)
• Motion blur
• Gaussian blur

SmartDeblur uses the FFTW library to provide its fast fourier tranformation implementation.
See fftw.org for details. All source files are under the GPL v3 license.

Author: Vladimir Yuzhikov ([email protected])

Detailed description of algorithms and methods you can see at my articles:
Restoration of defocused and blurred images (part 1) - Theory
Restoration of defocused and blurred images (part 2) - Practice

Sceenshots

Main Window:

Features

• High speed. Processing a 2048*1500 pixel image takes about 300ms in the Preview mode (when adjustment sliders can move). But high-quality processing may take a few minutes
• Real-time parameters changes applying (without any preview button)
• Full resolution processing (without small preview window)
• Deep tuning of kernel parameters
• Easy and friendly user interface
• Help screen with image example
• Deconvolution methods: Wiener, Tikhonov, Total Variation prior

Download

Version 1.27:
SmartDeblur-1.27-win.zip
SmartDeblur-1.27-mac.dmg

License

SmartDeblur is provided under the GPL v3 license

smartdeblur's Issues

SMARTDEBLUR 1.27 CRASHING ON WINDOWS

Hi Vladimir,

Hope you doing good.
Here is my feedback upon using SmartDeblur V1.27.

I downloaded the zip file - "SmartDeblur-1.27-win.zip — SmartDeblur - last windows version
7.0MB · Uploaded a month ago", from the site , https://github.com/Y-Vladimir/SmartDeblur/downloads.

The package is downloading and installing fine. After this I am extracting the files in the same directory. And running the SmartDeblur.exe. This resulting in SmartDeblur v.1.27 launching.

However, problem starting after this. The moment, I am uploading a .jpeg image file taken by a Canon 1000D camera, the image is loading in SmartDeblur and within seconds the SmartDeblur v.1.27 application is crashing

I removed the entire application, restarted my laptop, redownload the application all over again and
followed the same procedure. Still the same problem remains. Can you please help.

Please note : My laptop has Windows XP Professional Vrsion 2002 SP3 operating system, Intel Celeron 1.6GHz procesor and 1.24GB of RAM.

Thank you,

Kaushik Bhowmik,
From India.
[email protected]

UI broken on Linux

The UI looks wrong on Linux due to lack of a grid layout on the form.

Here is an example of what I'm talking about and a screenshot of a quick fix for just one problematic area:

Would you like a patch for all of UI?

BTW, what about using the slider widget from Kdenlive? You get both a label, a direct numeric input and a slider. Plus everything is perfectly aligned.

About screen not DPI Aware

My DPI level 125%, and the text is cut off and does not lay out right. Example:

Advanced opportunities to remove real motion blur

Hey Vladimir Yuzhikov!

First off all: Your tool is really amazing! From a photographers point of view it is already stunning what this tool can do to bad images.

However when it comes to removing motion blur, the real motion blur is many times very complex. Imagine shaking the camera to the left and afterwards up. To remove this blur, a kernel looking like a "L" would be required. Thus the user is not able to represent the kernel only using the sliders.

To solve this problem, I suggest creating an option to load a premade kernel. This kernel can be obtained by the user by using a image editing program (e.g. by extracting a bright spot on a dark background).

I'd be great if you could think about adding this new feature, because it would make your software so much more powerfull.

Greetings from Germany,

Rasenmaeher

Photo resolution limit

Hello,

I really like your program and found it useful for small photos. However, I guess you realise that modern cameras produce photos much larger than 3000 px on the long side. I tried opening a 24 MP photo from my Pentax K3-II but the program crashes. I am running it on a computer with 16 GB of RAM, so I do not see how a photo cannot fit into it (I assume that the crash is caused by a segmentation fault).

Also, I agree that processing large photos takes a long computational time but I think it will be worth it if we can make a crop of the photo, adjust the settings to make them pleasing to the eye and then plug in the full-size photo and let the program run in the background. It does not really matter if it takes half an hour if the result is good.

In any case, very nice work, I appreciate it.

I run the software on windows, as soon as the radius is set, the software crash suddently. Is there restrictions on the version of windows system? Thanks

My windows system is : windows 10, enterprise

Deblur challenge

What's the best you can make out of this heavily blurred picture ?
https://cdn.newsapi.com.au/image/v1/ad67fdaa76a11bd36e8195ce5d863789?width=1920

It's supposed to be a scan of 16mm film from a bolex camera with 240mm lenses (incorrectly mounted some say) exposed to a very bright object at a distance. Quite a challenge :)

Hope it's not offtopic.

I have a question

Did you send a paper on this technology? I want to quote.
Looking forward to reply

How do I set up the size of blur track?

How do I set up the size of blur track?

Qt5 build does not work on Gentoo

Hi,
I'm unable to build SmartDeblur on my Gentoo system which has only Qt5 (no Qt4 anymore).
I get
CMake Error at /usr/share/cmake/Modules/FindQt4.cmake:1323 (message):
Found unsuitable Qt version "5.12.0" from /usr/bin/qmake, this code
requires Qt 4.x

Many thanks for a hint,
Helmut

Program crashed under Linux Mint 13

At High-Quality processing I click to "Help" and then "OK". Program closes immediately without any messages.

Why do you give up the project?

I find the last update of this project is in 3 years ago, but why?

Semiblind deconvolution

Algorithm could get some more data after blind deconvolution by user input. I.e. if one has a blurred image of a screenshot and deblurs it (with a lot of additional moire), now he has a chance to tell that some area has the same color, or that some area doesn't have a strong gradient in it. The idea is quite straightforward, so there should be an article on it somewhere.

(If you're not interested in searching for possibly non-existing article or implementing such algorithm, please, leave this feature request opened in case someone likes to try it and make a pull request.)