Tuesday, April 27, 2021

Topaz Had The Sensor Format Argument 'On The Ropes' But Now DXO May Have The Sensor Format War 'Down For The Count!'

This file was processed in DXO PureRAW.  It was made with an Olympus E-M1 Mark III with the 
Olympus 12-100mm f/4 PRO lens @ 12mm; 1/13th sec. @ f/8; ISO 6400 (click to enlarge)
Below is a severe crop showing before and after being processed in DXO PureRAW.

It’s almost over.  The sensor size and camera format war, raging since the inception of digital photography, may soon be coming to an end.  What has made the difference?  Technology.  Technology is the great 'equalizer.'

In the beginning, there were huge differences in image quality between cameras with larger sensors, namely what is called medium format and full frame, and cameras with smaller sensors, i.e., APS-C, micro4/3 and even smaller.  Over the years, those differences have incrementally been diminished with improvements in technology.  Those improvements have come in the form of improved sensor technology, better in-camera image processing algorithms and amazing jumps in post image capture editing software technology.  A comparison of the differences between the digital cameras of the year 2000 to the cameras of today is nothing short of remarkable.

My hat is off to the sensor manufacturers for the improvements they have made in sensor technology (mainly Sony and Canon) as well as to the camera manufacturers for improvements in their in-camera image processor algorithms.  We've come a long way.

However, just as significant, if not more, arrived when Topaz Labs introduced their artificial intelligence (AI) driven plug-ins, specifically Gigapixel AI, Sharpen AI and DeNoise AI.  Those three particular plug-ins really closed the gap in image quality differences between the sensor sizes.  Image files could be increased in size with no apparent loss of quality.  Images that were 'just' not quite sharp can now be sharpened up nicely and high ISO images from smaller format sensor cameras can eliminate noise just as well as those from full frame cameras.

Gigapixel AI allows one to take an image made with a relatively low (by today’s standards) pixel count, such as micro4/3 20.2mp files, and upsize those to equal or surpass file sizes in cameras with the larger 36mp, 45mp, 50mp and even 60mp.  It is a rather wondrous piece of software.  The other benefit of this technology that I’ve used is to allow me to crop in significantly and then resize my final image appropriately as needed.  Whether you know it or not, you have seen many of those images in this blog.  I bet you didn’t notice any differences between those and ones that had not been resized.  Point made.

This crop is less than 1mp from the 20.2mp file.  This is how the file came out of the camera with
no noise reduction nor sharpening applied. (click to enlarge)

Topaz then released DeNoise AI and Sharpen AI.  Two amazing programs that can take an image in which you just missed precise focus or had a bit of camera shake, etc. and almost miraculously ‘fixes’ it.  It doesn’t work on every image but almost all of the ones I’ve tried.  I’ve used it many times on images that were not publishable and it the software corrected (usually my mistakes) and allowed me to illustrate this blog or make enlargements for framing and display.  Again, unless I pointed it out you probably didn’t notice.  Again, that’s the point.

Finally, DXO just released PureRAW.  All I can say is wow!  DXO is known for a few things.  The company is known for their terrific editing software, DXO PhotoLab 4, which is also a RAW converter, their Deep Prime noise reduction algorithms, in my opinion the best camera/lens correction modules in the industry and, of course, the now own and have updated the old NIK plug-ins that are revered by many.

With PureRAW, DXO gives us a standalone piece of software that very simply looks at your RAW image files, processes them to bring out detail (and I think it adds a bit of very clever sharpening), applies the Deep Prime noise reduction as well as the lens corrections.  The software allows you to keep your image files as RAW files in the form of DNGs or you can convert them to JPEGs.  The result is nothing less than spectacular, in my tests.  Another nice touch is that you don’t lose your original RAW files.  PureRAW asks you if you would like to export them to your editor along with the newly created DNGs or not not include them.  

Something that I noticed.  DXO shows you more of your original frame than does Lightroom Classic.  I didn't realize this before, but evidently LR Classic slightly crops all of the images it imports.  PureRAW shows you the entire image.  Since I discovered this, I'm bothered that LR crops my images as I am one of those individuals who precisely composes my images.  I wonder why Adobe does that?  Obviously, if DXO doesn't, neither does Adobe.

One last thing that I found.  When I copied some of my RAW files from my Lightroom Classic (LR) catalog into DXO PureRAW, I also copied the XMP Sidecar files created by LR, which contains all of my LR edits in a particular file.  After importing those into Lightroom, those previous edits were automatically applied and inserted into the DNG file. When the newly created DNG appeared in LR, it was fully edited.  That saves time in not having to re-edit a RAW file again.  Nice.

This is an identical crop from the same RAW file after it was processed in DXO PureRAW
(click to enlarge)

I chose a variety of micro4/3 files, from ISO 800 up to ISO 6400, processed them in DXO PureRAW then also processed them in Topaz DeNoise AI and LR, then compared the three versions.  In every case, the DXO software outperformed the other two.  I say that as the Topaz software also did an excellent job but on a few files, it introduced some tiny artifacts. One caveat—I only used Topaz DeNoise AI in its ‘automatic’ settings and didn’t adjust the sliders.  When I did spend some additional time to work with the sliders, I was able to equal the DXO output.  The difference between the two is time, effort and ease of use.

But it’s not all roses with the DXO software.  I found a few shortcomings.  First, it is not a plug-in so you have to run your files through it, then import them into your editing software. That means a significant change in your existing workflow.  

Second, when converting my RAW files to DNGs  for export into LR, they became more than three times the size of the original RAW files.  On average, a 16mp m4/3 RAW file  made with my old Olympus E-M5 or E-M1, with an average files size of 15.3mb turned into a DNG file with an average size of 55.9mb!  Twenty megapixel m4/3 files jumped from 19mb to an average of 68.8mb!  The losslessly compressed image files from my Nikon Z7 jumped from 61.5mb to a whopping 156.4mb!  Almost three times the size!  Now, this is not unheard of as other RAW to DNG converters with which I have experimented have resulted in about the same numbers.  However, you need to know this.  For me, instead of automatically sending all of my files through DXO PureRAW, I will most likely choose to send only select files as sending all of them through would require me to quickly buy a new primary hard drive as well as three new backup hard drives.  That's not an expense I need right now.

Third, even though the program will carry over your prior editing settings (from Lightroom, at least), the color and brightness of the files have been changed.  If your experience is similar to mine, you will have to do some adjusting to your processed files.

Fourth, converting each file takes quite some time.  My m4/3 files, on average, took 24 seconds and my 45mp Nikon files each, on average, took 56 seconds.  Now, I don’t have a very new computer.  Mine has a 4th generation Intel processor and Intel is currently selling computers with 11th generation processors.  So, if your computer is newer, with a fast GPU, your times may be considerably shorter.

Finally, DXO does not work with Fujifilm X files from the APS-C cameras.  The files are not recognized.  None of their software does.  As for the medium format cameras which don’t use the X-Trans sensor array, I don’t know if the software will or will not work with those.   If you have knowledge, please share it in a comment.

If you still think you need a large, expensive sensor in a large expensive full frame camera to make excellent images, you are behind the times.  As I’ve said many times in the past, for most of us under most all circumstances, a smaller format sensor camera will exceed all of our needs.  It is only when you are out photographing the extremes—extremely low light, extremely high contrast, need extremely high resolution, need extremely large enlargements, etc., is when a larger sensor with more pixels makes a difference.  However, the need for those extremes is being slowly (mostly) eliminated.

Technology is the great equalizer and it is only going to get better with time.  I find, when reviewing my images, that there really aren’t any I now make under any condition that I cannot make with my smaller format sensor cameras and lenses.  The value proposition to us photographers using smaller sensor cameras is great in the fact that smaller sensor cameras are normally less expensive, the lenses are less expensive and much smaller, the included features in the cameras are greater and the differences in fast, sure focus is becoming narrowed every year.  The technology for fast focusing, precise eye focusing, tracking, animal or bird focusing exists.  If it is not in a particular camera, that is the choice of the manufacturer, not a fault of a camera with a smaller sensor.  The manufacturers can put that technology in any camera with any sensor size.

Technology has become the great equalizer in many fields and among many products.  Whether through artificial intelligence, machine learning, the internet of 'things,' or just incremental advances, the format war in digital photography is becoming yesterday's news.  I'm convinced it is only through Internet and YouTube 'influencers' as well as the motivation of the camera companies to push us toward full frame (re: more profits in those larger format cameras).  The result is that most of us aren't convinced that smaller sensor cameras will serve almost all of our needs.  That is a shame.

I can’t speak for any of you, but these improvements excite me and I can’t wait for the cameras and technology of the future to arrive!  They will allow us to continue creating better and better images and open new doors to create images never before possible.

Topaz Labs has the sensor war 'on the ropes' and now DXO has it 'down for the count.'  Will the sensor war be knocked out anytime soon?  Probably not, but not for any logical reason.

Join me over at my website, https://www.dennismook.com 

Thanks for looking. Enjoy!  

Dennis A. Mook  

All content on this blog is © 2013-2021 Dennis A. Mook. All Rights Reserved. Feel free to point to this blog from your website with full attribution. Permission may be granted for commercial use. Please contact Mr. Mook to discuss permission to reproduce the blog posts and/or images.

7 comments:

  1. Dennis,
    Did I read correctly that you dragged both the RAW and the xmp file for an image already processed to some degree in LR and PureRAW sent back a dmg file with the LR adjustments already applied? I had understood that one had to copy the adjustments from the processed RAW and paste into the dmg in LR. If I read this correctly, a great discovery and real time/effort saver. BTW, agree with your thoughts :)

    ReplyDelete
    Replies
    1. Rick, you are correct. What I did was identify RAW images by ISO already in my Lightroom Classic catalog then, right clicking on the image (I have a Windows machine), then clicking “Show in Explorer” in the drop down box, I was taken directly to my identified image file in Explorer which appeared on my monitor highlighted in blue. I then copied the image file along with the XMP Sidecar file and pasted it in a newly created folder in which I was placing all of my test images.

      After accumulating a number of images of various ISOs, I opened PureRAW, imported the images (the XMP files are not visible in PureRAW), processed them and then exported the processed files to Lightroom. After importing in Lightroom, the DXO processed files automatically applied the edit data found in the XMP files.

      One thing I found that I mentioned. DXO changes the files color balance and density a bit so you may have to tweak them to get them where you want them. At least on my machine they did.

      That’s it. Hope this helps.

      Delete
  2. Dennis, thanks for this. I may yet go back to the OMD's. Really did like them a lot. If I read the reply to SW Rick correctly, the XMP files remained in the original folder after you copied the files to the test folder. Is it possible that Lightroom simply picked up the original .xmp files? I'm trying to figure out if DXO actually processes those .XMP files that Lightroom generated. Hope that makes sense.

    ReplyDelete
    Replies
    1. Rudiger, actually I copied both the image file me XMP file to PureRAW. That said, when looking at your files inside the PureRAW program, the XMP files were not visible. My understanding is that when PureRAW processes and converts a RAW file into a DNG, it inserts the contents of the XMP file into the DNG. So when importing into LR, the edits you previously made are embedded into the DNG.

      Delete
    2. Dennis... just an FYI. I also shoot with Leica, and DXO has very limited support for Leica cameras. Likely because they have no noise or lens correction required! (Just kidding). Topaz does a fine job with Leica files.

      Delete
    3. Rudiger, thanks. That is interesting. I thought that Fujifilm was the only system that DXO didn't support due to the X-Trans sensor. I wonder if it is because Leica uses the generic DNG as its RAW format?

      Delete