On 10/13/2013 05:23 PM, Patrick Shanahan wrote:
> * Max Killer <[email protected]> [10-13-13 03:58]:
>> On 10/10/2013 11:20 PM, Patrick Shanahan wrote:
>>> * Max Killer <[email protected]> [10-10-13 14:45]:
>>>> On 10/08/2013 05:16 PM, Patrick Shanahan wrote:
>>>>> Presently running darktable-1.3_1107_g83b15db_git-2.1.TM.x86_64.
>>>>>
>>>>> Adjusting denoise profiled strength, image displayed does not match
>>>>> exported image.  Exported image (jpg) shows much more noise at the same
>>>>> display size.  Makes it very difficult and time consuming to adjust noise
>>>>> strength level to match desired viewing size for best quality.
>>>>>
>>>>> Or am I doing something wrong?
>>>>> Screen size is 1920x1200
>>>>>
>>>>> see screen dump @
>>>>>   http://wahoo.no-ip.org/~pat/DarktableScreenshot.jpg
>>>>> and jpg exported @
>>>>>   http://wahoo.no-ip.org/~pat/130924_201112_dsd_7131_dt.jpg
>>>>>
>>>>> tks,
>>>> Could you please upload a file for inspection together with the xmp? So
>>>> we could have a look at the file and the output and try to reproduce....
>>> the original nef and the xmp file are at:
>>>     http://wahoo.no-ip.org/~pat/130924_201112_dsd_7131.nef
>>>     http://wahoo.no-ip.org/~pat/130924_201112_dsd_7131.nef.xmp
>>>     
>>> note that I do not observe or have not observed the same situation with
>>> today's issue:  darktable-1.3_1111_g4995137_git-2.1.TM.x86_64
>   [...]
>
>> the effect you observed is still present in the current git version, and
>> I would be surprised if not.
>>
>> You were comparing two different things in your first email. You
>> compared a "quick" resized image (dt viewport) with a full format
>> export. What you can compare is the quick resized view to a same size
>> export interpolated with the same interpolator.
>>
>> To make it short, if you _disable_ the high quality export and export
>> the image to, lets say 1500x1500, the output is more or less the same.
>> More or less because I didn't analyze it on the pixel level, but it
>> looks the same. Please be aware that the image viewer you are using is
>> also important. Eog for example uses another interpolator than geeqie
>> (which I use mainly).
>>
>> Now to the conclusion:
>> 1. Develop the images for the size you need them. If you need the full
>> res images, check the view in 100% mode regularly. The zoomed out view
>> uses an interpolator which can change the overall view.
>> 2. "Play" with the options, for example full res output. This also
>> depends on the iops used. Using high res output with your denoise
>> settings shows a grainier image.
>>
>> As a side effect, the image now exports very fast...
> But what is the difference in the output image if export is not set to
> "high quality", particularly the printed output?
>
> tks much for explanation.
Hello,

first, I don't understand your point with the "printed output".

The difference is the order, in which the image operations are performed:
If high quality export is off, the the image is resized first, and then 
the resized image is used as input for all your selected iops.
If you turn the high quality export on, the full resolution is used as 
the input for the iops and the image is scaled down afterwards. This is 
the reason why the export takes much longer compared to low res output 
if a smaller resolution is used, the pipe has simply to calculate a lot 
more pixels.

But as some iops are "dependend" on the image size, the output has to be 
different.
If you export to full resolution, then the high/low resolution option 
has no effect, because no scalling takes place.


hal

------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from 
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
_______________________________________________
Darktable-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/darktable-users

Reply via email to