My experience with DXO vs. DPP vs. no noise reduction

[freelanceshots]

I downloaded the trial version of DXO pro and thought I'd test the differences between it and DPP. I took a RAW image I shot with my 40D which has more noise in certain shots and processed the images 2 ways. First I used DPP to process the image with no luminance or chrominance noise reduction where I left sharpness as shot which is like a plus 4 and a -2 on color saturation. Next I took the same image and processed it but used number 5 luminance and 2 chrominance RAW noise reduction.


In the end I have 3 images one with no noise reduction, one with the DPP reduction that I felt comfortable with and then the no noise reduction image that I then precessed in DXO.


I did not do any specialized things in DXO as I am new to it but I did take some of the preset out of the basic workspace and just selected the corrections for distortion, unsharp mask, noise and chromatic aberration. Images where then loaded into PS CS5 and cropped down about 30 percent at 100 dpi 1500px x 1000px and uploaded to flickr.


What is everyone else's thoughts? Is there something that i did not do correctly or needs to be redone? If in fact DXO does handle noise better then I will most likely buy it down the road but I'm not sure that there is a noticeable difference in my test. I'll keep looking at it for a couple of days and maybe I will have a different opinion.





Images taken off and replaced with new ones below.

[Daniel Browning]

If I were going to compare noise reduction, I would do this:


* Choose a part of the image that was so noisy that it really needs noise reduction.


* And has very high detail (noise reduction is sometimes a balance between detail and noise).


* View at 100% crop.


A 1500 pixel image roughly corresponds to the detail visible a 4x6 print. If that

[freelanceshots]

OK scratch the above images. I don't do a lot of printing where most of my work is done for online. I'm interested in noise reduction that will help my work mainly in this area. I don't use noise reduction very much with the 5D2 but when I do I want to get the best results with the least amount of work. I took your advice Daniel and chose another image shot with the 40D which usually has more noise. This was one of many inventory product shots that i did a few days ago where I had to make them look a lot better then they came out of the camera. I used my 40D this time because I never use it and wanted to give my 5D a break. I never noticed how far off my 40D's display was before but these images where way off in brightness and white balance. Anyways, here are my results from CS5, DPP, DXO, and no noise reduction. I had them at 300dpi and approx 100 percent crop but I shot the images at small resolution so 1936 x 1288px is the original size. Same principal applies from my first post where I was adjusting for nice, quick results using either CS5, DXO and DPP.


[View:http://community.the-digital-picture...neric/utility/ original image by freelanceshots, on Flickr:550:0]


[View:http://community.the-digital-picture...neric/utility/ CS5 reduction by freelanceshots, on Flickr:550:0]


[View:http://community.the-digital-picture...neric/utility/ DXO reduction by freelanceshots, on Flickr:550:0]


[View:http://community.the-digital-picture...neric/utility/ DPP reduction by freelanceshots, on Flickr:550:0]


[View:http://community.the-digital-picture...neric/utility/ no noise reduction by freelanceshots, on Flickr:550:0]

[freelanceshots]

DXO results look the best to me with sharper edges and more contrast.

[neuroanatomist]

DXO results look the best to me with sharper edges and more contrast.

Agreed. Some time back, I compared DPP vs. DxO and found that DxO did a better job with NR and other corrections.

[William]

DXO results look the best to me with sharper edges and more contrast.

Yes that was my conclusion too.

[Joel Eade]

Try downloading a trial of the Topaz Denoise plugin and compare it with DXO pro.


It is less expensive and seems to work better than CS5 or DPP for me.

[freelanceshots]

Here is another test that I did today just to see what I'd find different with a new image shot at ISO 2000 with 5D2 and the 100mm IS macro lens shutter 1/10, f/3.2.


These images where shot RAW so the final crop images are larger. On both of my examples I did no upscaling in photoshop as these are the crops at 300dpi saved as tif's until the flickr upload which was jpg.


[View:http://community.the-digital-picture...neric/utility/ no crop reduction test by freelanceshots, on Flickr:550:0]


[View:http://community.the-digital-picture...neric/utility/ CS5 crop noise reduce by freelanceshots, on Flickr:550:0]


[View:http://community.the-digital-picture...neric/utility/ DPP crop noise reduction by freelanceshots, on Flickr:550:0]


[View:http://community.the-digital-picture...neric/utility/ DXO crop noise reduction by freelanceshots, on Flickr:550:0]


[View:http://community.the-digital-picture...neric/utility/ crop no noise reduction by freelanceshots, on Flickr:550:0]

[freelanceshots]

I don

[Fast Glass]

Looks like the second from last is the best, it has the most dietail and much finer grained noise compared to DPP. The one with CS5 is very close andbut I wouldstill give a slight edge to DxO, but it's very good also.


Here is a situation I routinely come across in birding,soft lighting but high ISO.


[img]/cfs-file.ashx/__key/CommunityServer-Discussions-Components-Files/24/6685.IMG_5F00_0555-100_2500_-crop.JPG[/img]


[img]/cfs-file.ashx/__key/CommunityServer-Discussions-Components-Files/24/5025.IMG_5F00_0555_5F00_DxO-TIFF-100_2500_-crop.JPG[/img]


ISO 3200 pushed one stop, 100% crop. I think the difference is clear, but I won't tell just yet! Try to guess which is DPP and DxO.


In higher contrast scenes with deeper shadows were the noise can be even worse the difference can be even more, but I choose this expample because it's closer to what I or other people would shoot in real life. Butif youhave an important image it's nice to have better noise reduction. I have a light hand in noise reduction and this expample could probably NR a little more if you wanted.


Cheers,


John.