Raw processing comparison: DxO PhotoLab vs C1

Yoms · April 12, 2021, 1:03am

EDITED: added comparison with linear DNG exported from DxO to C1

Hi,

I’ve made a raw processing comparison for myself and thought it could interest some of you. So I share it here. The tests focus on details extraction, shadows and highlights recovery.

Here’s the link to download the screenshots I’ve made:

I’m only interested in Lightroom Classic as a DAM. That’s why you won’t find anything in that regards.

All explanations are provided in the readme 1st.txt file. I copy-paste the content of this file below for those of you who want to read it first (pun intended) before downloading the screenshots:

RAW PROCESSING / DEMOSAICING COMPARISON

Capture 1 21 Pro build 14.1.0.220 is on the left / DxO PhotoLab v4.2.0 build 51 on the right on each picture.

Images 1 to 10: Canon 5D Mark III + Canon EF 85mm f/1.8 USM @ f/11
Images 11 and 12: Canon 5D Mark III + Canon EF 17-40mm f/4L USM @17mm f/5.6


Images 1 & 4: how much detail can you extract from plain raw files?
- C1's default unsharp mask reset to "No sharpening"
- No optics modules enabled for both C1 and DxO PL


Images 2 & 5: how much "added" detail can you get by enabling optics modules corrections?
- Optics modules enabled for both C1 and DxO PL (all corrections enabled, incl. "Lens sharpness")
- C1's default unsharp mask is STILL reset to "No sharpening"


Images 3 & 6: same concept as 2 but...
- C1's default unsharp mask is now enabled


Images 7 & 8: highlight recuperation tests
- Optics modules enabled for both C1 and DxO PL (all corrections enabled, including "Lens sharpness")
- C1's default unsharp mask reset to "No sharpening"
- Highlight is the *only* cursor/slider moved for this test (apart from optical corrections)


Images 9 & 11: how dark areas are handled by default & how much detail can you extract from these areas without moving the shadow recovery cursor/slider?
- Optics modules enabled for both C1 and DxO PL (all corrections enabled, including "Lens sharpness")
- C1's default unsharp mask reset to "No sharpening"


Images 10 & 12: shadow recovery tests
- Optics modules enabled for both C1 and DxO PL (all corrections enabled, including "Lens sharpness")
- C1's default unsharp mask reset to "No sharpening"
- Shadow is the *only* cursor/slider moved for this test (apart from optical corrections)


Images 13: how dark areas are handled by default from a DxO export to C1 in linear DNG?
- Optics modules enabled for DxO PL only (all corrections enabled, including "Lens sharpness"), then from that result a linear DNG is exported to C1 
- C1 optics module correction *disabled* and default unsharp mask reset to "No sharpening"


Images 14: shadow recovery tests from a Dx0 export to C1 in linear DNG
- Optics modules enabled for DxO PL only (all corrections enabled, including "Lens sharpness"), then from that result a linear DNG is exported to C1 
- C1 optics module correction *disabled* and default unsharp mask reset to "No sharpening"
- Shadow is the *only* cursor/slider moved for this test (apart from optical corrections)



DON'T READ THE FOLLOWING BEFORE MAKING YOUR OWN MIND.

Analysis:
1) By default, i.e without one single correction enabled, DxO PL renders more contrasty and darker images than C1, but not sharper images,

2) When enabled, DxO PL's optics processing - especially "Lens sharpness" correction - is pretty effective; C1 has to enable both the same correction *and* default unsharp mask to match DxO PL "Lens sharpness" correction only (and unsharp mask is not even enabled on DxO PL at this stage),

3) DxO PL has more highlights recuperation range; when C1 is at -100, DxO provides the same result at -65 or even -35 on these examples,

4) But, pushing HL recuperation closer to -100 on DxO (not shown here) results in ugly colour shift, unlike C1 who handles this aspect perfectly, albeit reaching the -100 limit earlier; so despite the greater range, DxO HL recuperation is not better, it's a draw between both apps,

5) C1 really puts DxO PL to dust in terms of both default shadow rendering (w/o adjustments) and shadow recovery; this is coherent with first bullet that stated that "DxO PL renders more contrasty and darker images than C1",

6) When using the shadow recovery cursor, DxO generates noisier results; shadow recovery might be considered as a draw between both apps by some if you add DxO PL's superior noise reduction in the equation),

7) Linear DNG test files tell us that, despite being first demosaiced by DxO, C1's raw engine still performs the same way as regards shadow recovery. In other words, it's not a prerequisite to demosaic the file internally to get excellent shadow recovery; C1's shadow recovery works just as well with a linear DNG file.

mwsilvers · April 12, 2021, 2:46am

@Yoms,

With regard to your comments about shadow recovery, I rarely use the PhotoLab global shadows slider for that purpose since its range extends too far into the midtones. I much prefer using local adjustments for applying shadow detail recovery exactly when it is needed without impacting the overall image. I then finish up with DeepPRiME which eliminates noise from the recovered shadows areas.

Mark .

platypus · April 12, 2021, 10:33am

Your findings correspond to what I see when I compare DPL, C1 (and Lr)

My lessons learned from such comparisons - and your LL might be different - are the following:

All tools have strong and weak points
No tool is best in absolute terms (use case, features, ergonomics, usability etc. vs. requirements)
Work with and stick to the tool that best meets your requirements
Use other tools only if beneficial

Yoms · April 12, 2021, 4:46pm

Thanks for your input. I understand what you’re saying, but the purpose was solely to compare raw engines: how much detail can be extracted + handling of highlights and shadows. That’s the most commonly sought for characteristics of raw engines.

What you describe is rather a workflow which is absolutely fine, but it was not my initial intent. That said, I made an allusion to that workflow as well in my own analysis when I wrote:
"shadow recovery might be considered as a draw between both apps by some if you add DxO PL’s superior noise reduction in the equation"

Yoms · April 14, 2021, 10:05pm

Hey,

I’ve just added a new comparison to the files. First post and readme 1st.txt have been edited accordingly.

The idea was to see if C1’s default handling of dark areas and shadow recovery were performing the same way when given a linear DNG file demosaiced by DxO. This way, you can have both the optics and noise corrections from DxO while still handling shadow recovery in C1.

This is somewhat C1 + DxO PureRaw.