Troubles with AutoSelection of AutoProfile area ...

questions about practical use of Neat Image
Post Reply
mhall
Posts: 1
Joined: Fri Jun 06, 2008 2:31 am

Troubles with AutoSelection of AutoProfile area ...

Post by mhall »

Hello,

I've been using NeatImage for nearly five years now and I'm very happy with the steady increase of speed and quality that has been incorporated into the software. Overall I consider myself very proficient with the software and have integrated it into my workflow software via the command line to automate the processing of the 10s of thousands of images we have to retouch each year.

I photograph indoor volleyball extensively and each event sees us commonly processing thousands of images photographed at anywhere from ISO 1600 - 6400. Often with excellent exposure, but under those fun convention center lights which fluctuate in color and exposure between each frame, making a standard noise profile impractical due to constant changes not just in exposure but WB, both of which are corrected in the RAW conversion stage.

I use Bibble to convert my RAW files and have purchased the version of NoiseNinja to possibly integrate my noise reduction straight into RAW conversion, but find that I MUCH prefer NeatImage's interface and results.

However, I have noted that NI's selection of the area to be auto profiled is heavily biased to the highlights. It has been so since the feature was added and I've never been able to make sense of this. Noise is minimal to low in the highlights and anytime NI selects a highlight to profile, I get a bad image as a result, which I need to manually reprocess. With convention center lights splashing onto background walls, it is not uncommon that I get a large highlight region in the background of an image that causes the AutoSelection function to choose poorly.

Is this simply because NI is mistaking the highlight region for smooth featureless regions and mistaking midtone and shadow areas of no detail (but with noise) areas of high frequency noise?

The problem I am running into is that having to stop my processing workflow to go back and reprofile and process random images throughout an events worth of work is very time consuming and disruptive to the processing workflow.

Is there any way to bias the tonality of the region NI looks to AutoProfile?

Thanks for great software.

Regards,
Micheal
NITeam
Posts: 3173
Joined: Sat Feb 01, 2003 4:43 pm
Contact:

Post by NITeam »

Hello Micheal,

Thank you for your kind comments and questions.

In principle, it is possible to bias the selection but that has to be done in the program itself or there should additional controls for that. We currently work on some improvements to that part of Auto Profile to make NI select better areas in automatic mode. I think it will be possible to improve the performance of Auto Profile on such images. BTW, you may want to send some typical examples of those images to help us ensure the modified AP works well with them.

Also, even if there is a fluctuation of color and exposure, it still makes sense to use pre-built profiles (especially if you have built them for your camera usign the Calibration Target workflow) complemented by Auto Fine-Tune. Auto Fine-Tune will compensate for those fluctuations so that the loaded and fine-tuned profile will in most cases be as good as a profile built using Auto Profile with manual inspection of the selected area.

Hope this helps.
Vlad
Post Reply