Re: What does the 'F' mode mean in PIL?

A followup, trying to do something real

It seems that the raw decoder doesn't understand all the modes listed in:

http://www.pythonware.com/library/pil/handbook/decoder.htm

indeed, hardly any of them ('F' only for the "floating point modes")

Enclosed is some sample code I've got that attempts to convert a 
unsigned 16 bit grayscale image to RGB. I can't get it to do anything, 
as I get:

ValueError: unknown raw mode

for any floating point mode other than 'F'

What am I missing here?

-Chirs


Christopher Barker wrote:
> Hi folks,
> 
> I've been looking through the docs on File Decoders (or, in my case for 
> decoding binary data with fromstring() or frombuffer()):
> 
> http://www.pythonware.com/library/pil/handbook/decoder.htm
> 
> With the raw decoder, most of the modes are clear to me, but the 'F' 
> mode has me confused:
> 
> What is it really storing, working with?
> - one floating point value per pixel?
> - one floating point value per colorband?
>   - but then, how are they arranged?
> - Is it single precision (32 bit) floats only?
> 
> 
> What does it mean if you use one of the modified 'F' modes, like 
> "F;16N"? I know what an 16 bit native unsigned integer is, but I don't 
> understand:
> 
>  - What happens if I do fromstring() with a mode f "F;16N" ?
>   - is the data converted to a float (32bit)?
>     - if so, do you just lose information if you have higher precision 
> types being passed in?
>   - is the raw data stored in 16 bit (two byte) chunks?
> 
> 
>  - How is 'F' mode interpreted if you convert to, say 'RGB' mode?
>    -grayscale?
>    - something else?
> 
> 
> Any enlightenment would be great.
> 
> -Chris
> 
> 


-- 
Christopher Barker, Ph.D.
Oceanographer

Emergency Response Division
NOAA/NOS/OR&R            (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

[email protected]

_______________________________________________
Image-SIG maillist  -  [email protected]
http://mail.python.org/mailman/listinfo/image-sig

#!/usr/bin/env python

from PIL import Image
import numpy as np

width, height = (20, 30)

max_value = np.iinfo(np.uint16).max
raw_data = np.random.randint(0, max_value, (width, height)).astype(np.uint16)

def rgb_from_int16_PIL(arr):
    """
    converts a 2-d array of uint16 grayscale values to a 24 bit RGB array,
    using PIL
    """
    #i = Image.fromstring("F;8", (width, height), arr)
    i = Image.fromstring('RGB', # mode
                         (width, height), # size
                         arr.tostring(), # data
                         "raw", 
                         'F;16', # raw mode
                         # stride
                         # orientation
        )

    return np.fromstring(i.tostring(), dtype=uint8).reshape((width, height, 3))

rgb_data = rgb_from_int16_PIL(raw_data)