GStreamer Exposure Compensation (Exposure Value) - gstreamer

I have an imaging source which has a bit depth of 12 feeding into GStreamer. The eventual output will have a bit depth of 8. The scene being imaged is very low light. I can take images captured from this source (16-bit TIFFs), and in GIMP I can adjust the Exposure (Colors -> Exposure) to make details visible which were not previously.
Currently, when GStreamer is converting from 12-bit to 8-bit, it seems like it is compressing the entire 12-bit range into the smaller 8-bit range. What I'd like to do instead is maybe chop off the upper 4 bits, or ideally apply some sort of level curve to the image (which is what I think GIMP is doing).
I've looked around, but can't seem to find any way to do this. I've tried the videobalance plugin, but it doesn't seem to do what I want.
Any ideas?

Related

Using OpenGL to perform video compositing with YUV color format - performance

I have written a C/C++ implementation of what I term a "compositor" (I come from a video background) to composite/overlay video/graphics on the top of a video source. My current compositor implementation is rather naive and there is room for CPU optimization improvements (ex: SIMD, threading, etc).
I've created a high-level diagram of what I am currently doing:
The diagram is self explanatory. Nonetheless, I'll elaborate on some of the constraints:
The main video always comes served in an 8-bit YUV 4:2:2 packed format
The secondary video (optional) will come served in either an 8-bit YUV 4:2:2 or YUVA 4:2:2:4 packed format.
The output from the overlay must come out in an 8-bit YUV 4:2:2 packed format
Some other bits of information:
The number of graphics inputs will vary; it may (or may not) be a constant value.
The colour format of the Graphics can be pinned to either ARGB or YUVA format (ie. I can provide it as you see fit). At the moment, I pin it to YUVA to keep a consistent colour format.
The potential of using OpenGL and accompanying shaders is rather appealing:
No need to reinvent the wheel (in terms of actually performing the composition)
The possibility of using GPU where available.
My concern with using OpenGL is performance. Looking around on the web, it is my understanding that a YUV surface would be converted to RGB internally; I would like to minimize the number of colour format conversions and ensure optimal performance. Without prior OpenGL experience, I hope someone can shed some light and suggest if I'm about to venture down the wrong path.
Perhaps my concern relating to performance is less of an issue when using a dedicated GPU? Do I need to consider separate code paths:
Hardware with GPU(s)
Hardware with only CPU(s)?
Additionally, am I going to struggle when I need to process 10-bit YUV?
You should be able to treat YUV as independent channels throughout. OpenGL shaders will be calling them r, g, and b, but it's just data that can be treated as whatever you want.
Most GPUs will support 10 bits per channel (+ 2 alpha bits). Various will support 16 bits per channel for all 4 channels but I'm a little rusty here so I have no idea how common support is for this. Not sure about the 4:2:2 data, but you can always treat it as 3 separate surfaces.
The number of graphics inputs will vary; it may (or may not) be a constant value.
This is something I'm a little less sure about. Shaders like this to be predictable. If your implementation allows you to add each input iteratively then you should be fine.
As an alternative suggestion, have you looked into OpenCL?

What is the raw form of a compressed image file format(jpeg, PNG , gif)?

As we know jpeg , PNG , gif are all compressed file formats, my question is what is the original source of input we provide to these compression algorithms and in which form a image data is stored before it gets converted into one of these file formats.
That depends.
PNG is generally lossless, but it does have a limit on the number of bits/pixel. GIF turns out to be lossless, too, but it is more complicated to get a high number of colors. These formats are still compressed, but use a compression that doesn't lose data.
JPEG is lossy. If you save as a JPEG, you will not be able to revert back to another format without losing some clarity. By representing the data as equations it can get quite small, but it can start to look "blurry" as the approximations get worse.
There are other images formats, like TIFF, RAW and BMP, which generally don't do any compression, although they are really more like containers and technically can contain compressed data, but they usually don't.
The original, uncompressed, data depends on what generates it. A photoshop file will save as a PSD but internally may represent it differently in memory. Every digital camera may have a different way of laying out its internal memory, and the photo sensors tend to map 1 to 1 from a sensor to a memory location of a set number of bits.
The common pattern, however, is that each pixel of the image is stored as 3 (sometimes 4) color values, each one between 8 and 16 bits. The 3 values may represent Red, Green and Blue, or alternatively Hue, Saturation and Value. For design, it could be CMYK (Cyan, Magenta, Yellow and blacK). There could also be an alpha value. It's unusual to use more than 16 bits for each color channel and most common to use 8. Using 12 bits is considered by most to be full color, but that doesn't align very well on 32 bit or even 64 bit machines. Still, 12 bit is used sometimes in digital video signals since when broadcast serially the color values don't need to fit into words.
Different formats will go in a different order. Usually rows first, but some formats start at the bottom row and some start at the top.
So, the real answer is it depends on what the particular compressor is looking for. Most software that saves as JPEG or PNG will accept multiple formats and the most common is probably 32bit/pixels with 8 bytes each for RGB (red, green, blue) and one either unused or alpha. It will need width and height of the image so the image data should be width*height*4 in bytes. You generally pass in a defined constant that tells it the byte order: RBGA, ARGB, BGR, RGB, etc.

How can I compress jpeg image with compression rate 4 bpp or less?

I am trying to compress my .jpeg image in Photoshop.
WHat is the best way to do this?
I am now calculating the bpp taking the image size in kb, calculating how many bits that is. Then I take the image size in pixel*pixel to get the amount of pixels in the image. After that I divide bits/pixels, to find how many bits per pixel the image has.
But How can I change this number? My guess is to change how many kb the image is, but how do i do this?
Thanks for any help!!
Yes, you can achieve higher compression ratio than 4 bits per pixel. Images with solid color can have rate as low as 0.13bpp.
In fact 4bpp is quite poor compression — it's same as uncompressed 16-color image or half of 256-color image, which even GIF can manage. JPEG can look decent at 1-2bpp.
in general, you cannot "compress" a jpeg image. all you can do is to reduce the image quality further in order to achieve a lower bpp value. jpeg streams are always compressed and they use a lossy compression method. it means that the original image will never ever be reconstructed from a jpeg image. the smaller the file the more information you have lost.
a specific "bpp value" is not, and should never be your target. especially with lossy compression. you should always look at your current image and decide whether it is still good enough or not.
if you still have the original image, try a lossless compression format, like zip-compressed or lzw-compressed tiff or compressed png. i'm sure PhotoShop can handle these formats as well. another softwares like IrfanView (https://www.irfanview.com/) or XnView MP (https://www.xnview.com/en/xnviewmp/) will convert your images too.
if you want manual (eg. full) control over your images, you should use command line utilities, like ImageMagick (https://imagemagick.org/) or NConvert (please find the XnView MP link above)
if you have only the jpeg images do not touch (edit & save) them. with every single save operation you lose another bunch of information. you should always work on file copies.
you should always keep your master image (the very picture you took with your phone or your camera).
of course, these rules of thumb will not answer your original question.

Writing 10,12 bit TIFF files with LibTIFF C++

I'm trying to write 10,12 bit RGB TIFF files with LibTIFF.
The pixel data is saved locally in an unsigned short buffer (16bits)
1) If I set TIFFTAG_BITSPERSAMPLE to 10 or 12, not enough bits are being read from the buffer, and the output is incorrect. (I understand that it is just reading 10 or 12 bits per component, instead of 16 and this is the problem)
2) I tried packing the bits in the buffer, so that it is really 12-R, 12-G, 12-B. In this case, I think the file is being written correctly but no viewer I could find could display this image properly.
3) If I set TIFFTAG_BITSPERSAMPLE to 16, viewers can display the TIFF image, but then I have a problem that I don't know if the image was originally 10 or 12 bits (If I want to later read it with LibTIFF). Also, the viewer expects the dynamic range to be 16 bits and not 10 or 12, also resulting in a bad view.
4) The most annoying part is that I couldn't find one 10, 12, or 14 bit TIFF image on the web to see what the header is supposed to look like.
So finally, what is the proper way to write 10 or 12 bit Image data to a TIFF file ?????
The TIFF specification does not specify a way to store 10, 12 or 14 bits per channel in an image. Depending on the encoder and decoder, it may still be possible to work with such images, but it is effectively an implementation detail, as they are not required to do this.
If you want more than 8 bits of precision in a TIFF, your only choice is 16 (or floating point, but that's a different story).
I'm not aware of any image format with specific support for these bitdepths, so viewers will likely be a problem anyway if you must store the image with that specific bitdepth. The simplest workaround I can think of would be to just store as 16 bits per pixel and put the original bitdepth as metadata (e.g. in an ImageDescription tag), but it all depends on what the images will be used for and why you need this information.
You can store the image as a multi-image file. For example, with a 12 bit source, one image would be an RGB(8) image using the upper 8 bits and a second 16bit gray scale that was a combination of the low four bits and four bits of padding. This gives a TIFF that can be viewed with on a monitor with standard programs and the extra precision can be retrieved with custom software.
I disagree that 'exotic' bit depths are not good. This format would reduce the image size by 5/6. You could even just store the 2nd image as a re-scaled version that would have the 4 bits tightly packed without padding for a 3/4 size reduction. This savings can be significant with very large data sets, where compression is not an option due to the nature of the data. Ie, many scientific and machine vision applications may want the un-adultered bits. The ability to convert from the multi-image tiff to a 16-bit tiff would allow the use of standard programs and image libraries.

image color conversion

I need to convert 24bppRGB to 16bppRGB, 8bppRGB, 4bppRGB, 8bpp grayscal and 4bpp grayscale. Any good link or other suggestions?
preferably using Windows/GDI+
[EDIT] speed is more critical than quality. source images are screenshots
[EDIT1] color conversion is required to minimize space
You're better off getting yourself a library, as others have suggested. Aside from ImageMagick, there are others, such as OpenCV. The benefits of leaving this to a library are:
Save yourself some time -- by cutting out dev and testing time for the algorithm
Speed. Most libraries out there are optimized to a level far greater than a standard developer (such as ourselves) could achieve
Standards compliance. There are many image formats out there, and using a library cuts the problem of standards compliance out of the equation.
If you're doing this yourself, then your problem can be divided into the following sub-problems:
Simple color quantization. As #Alf P. Steinbach pointed out, this is just "downscaling" the number of colors. RGB24 has 8 bits per R, G, B channels, each. For RGB16 you can do a number of conversions:
Equal number of bits for each of R, G, B. This typically means 4 or 5 bits each.
Favor the green channel (human eyes are more sensitive to green) and give it 6 bits. R and B get 5 bits.
You can even do the same thing for RGB24 to RGB8, but the results won't be as pretty as a palletized image:
4 bits green, 2 red, 2 blue.
3 bits green, 5 bits between red and blue
Palletization (indexed color). This is for going from RGB24 to RGB8 and RGB4. This is a hard problem to solve by yourself.
Color to grayscale conversion. Very easy. Convert your RGB24 to YUV' color space, and keep the Y' channel. That will give you 8bpp grayscale. If you want 4bpp grayscale, then you either quantize or do palletization.
Also be sure to check out chroma subsampling. Often, you can decrease the bitrate by a third without visible losses to image quality.
With that breakdown, you can divide and conquer. Problems 1 and 2 you can solve pretty quickly. That will allow you to see the quality you can get simply by doing coarser color quantization.
Whether or not you want to solve Problem 2 will depend on the result from above. You said that speed is more important, so if the quality of color quantization only is good enough, don't bother with palletization.
Finally, you never mentioned WHY you are doing this. If this is for reducing storage space, then you should be looking at image compression. Even lossless compression will give you better results than reducing the color depth alone.
EDIT
If you're set on using PNG as the final format, then your options are quite limited, because both RGB16 and RGB8 are not valid combinations in the PNG header.
So what this means is: regardless of bit depth, you will have to switch to index color if you want RGB color images below 24bpp (8 bits per channel). This means you will NOT be able to take advantage of the color quantization and chroma decimation that I mentioned above -- it's not supported in PNG. So this means you will have to solve Problem 2 -- palletization.
But before you think about that, some more questions:
What are the dimensions of your images?
What sort of ideal file-size are you after?
How close to that ideal file-size do you get with straight RBG24 + PNG compression?
What is the source of your images? You've mentioned screenshots, but since you're so concerned about disk space, I'm beginning to suspect that you might be dealing with image sequences (video). If this is so, then you could do better than PNG compression.
Oh, and if you're serious about doing things with PNG, then definitely have a look at this library.
Find your self a copy of the ImageMagick [sic] library. It's very configurable, so you can teach it about the details of some binary format that you need to process...
See: ImageMagick, which has a very practical license.
I received acceptable results (preliminary) by GDI+, v.1.1 that is shipped with Vista and Win7. It allows conversion to 16bpp (I used PixelFormat16bppRGB565) and to 8bpp and 4bpp using standard palettes. Better quality could be received by "optimal palette" - GDI+ would calculate optimal palette for each screenshot, but it's two times slower conversion. Grayscale was received by specifying simple custom palette, e.g. as demonstrated here, except that I didn't need to modify pixels manually, Bitmap::ConvertFormat() did it for me.
[EDIT] results were really acceptable until I decided to check the solution on WinXP. Surprisingly, Microsoft decided to not ship GDI+ v.1.1 (required for Bitmap::ConvertFormat) to WinXP. Nice move! So I continue researching...
[EDIT] had to reimplement this on clean GDI hardcoding palettes from GDI+