Questions and postings pertaining to the usage of ImageMagick regardless of the interface. This includes the command-line utilities, as well as the C and C++ APIs. Usage questions are like "How do I use ImageMagick to create drop shadows?".
The resulting image looks extremely blurry and much too dark.
We are using ImageMagick 6.8.3.
Is there any way to produce results similar to ImageMagick 6.7.6 without the use of color profiles?
We recently Upgraded our Version to produce TIFF images in gray colorspace (link).
In IM 6.7 the result looks like this. In IM 6.8 like that. The result in 6.7 is nearly the same like the source. In 6.8 the result does not match the source. Is it possible to get the same result in 6.8 like in 6.7?
EDIT: It seems like the sRGB->RGB conversion is the Problem. IM 6.7: rgb.jpg, IM 6.8: rgb.jpg
Yes, 6.7.6 was when IM was rather confused about RGB/sRGB.
If "in" is in colorspace sRGB, then "convert in -colorspace RGB out" should convert it to colorspace RGB. Approximately, this should convert values (from 0 to 1) by raising them to the power 2.2, which will lower them.
The source has overall mean 0.534136; raising to power 2.2 is 0.2485. This is much closer to the result you get from 6.8.3-9 than 6.7.6-10.
After doing the conversion, the pixel values are lowered, and the values represent RGB colorspace. Sadly, JPEG files don't seem to have a flag that says the values are RGB. So viewers think it is sRGB. As the values have been lowered, the image looks darker.
I haven't been following this thread because I don't often work in CMYK, but I had a collision of two thoughts that MAY give a useful direction of inquiry concerning the blurring.
Thought 1: Your command doesn't do anything inter-pixel, so even a bug couldn't readily cause blur, so this is probably a channel sub-sampling issue when saving. Incorrect sub-sampling could be easily overlooked in jpeg's default Y'CBCR, but internal RGB or CMYK would show sub-sampling as clearly visible blurring.
Thought 2: I remember glancing something unexpected concerning sub-sampling. I didn't look into it because my process is usually jpeg -> png -> jpeg and this only shows up for jpeg -> jpeg. The documentation says that the default sub-sampling is based on the -quality setting, but it also says that the input jpeg's settings will be used for the output unless changed. So when you explicitly set -quality but do not explicitly set -sampling-factor, which controls the sub-sampling for the output?
I am using Version 6.7.6-10 and Version 6.8.3-9 on Linux (x64).
versions between 6.7.6.7 and 6.7.7.7 were when things were unstable regarding IM color space conversions. I would not tend to believe results in that range.
Note that -quality should be specified after reading the input jpg and before writing the output jpg.
1. The OP mentions upgrading to make gray colorspace TIFFs. I suggest it is worth upgrading to 6.8.5, because that version (unlike 6.8.4, and versions back to 6.7.9) doesn't seem to make the assumption that greyscale images are RGB.
2. If any further processing it to be done to the images, I suggest avoiding JPG. The lossy compression used by JPG introduces artefacts that may be invisible in the JPG image, but become revealed by subsequent processing.
