Luminous Landscape Forum

Raw & Post Processing, Printing => Digital Image Processing => Topic started by: Bob Rockefeller on November 24, 2016, 08:56:22 am

Title: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on November 24, 2016, 08:56:22 am
With the pre-release of ON1 Photo RAW available, and that fact that it may have the first new RAW converter developed in years, has anyone with the necessary skills examined the quality of those conversions?

Adobe has been in the imaging business since forever and ACR is often the Pro's choice. Phase One is well respected for the quality of its RAW conversions. Is there something new under the sun? Could ON1 compete on its quality of RAW conversion?
Title: Re: Early ON1 Photo RAW testing?
Post by: Cem on November 24, 2016, 09:12:03 am
Take a look at this post (http://forum.luminous-landscape.com/index.php?topic=114584.msg945361#msg945361) Bob.
Title: Re: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on November 24, 2016, 09:14:59 am
Take a look at this post (http://forum.luminous-landscape.com/index.php?topic=114584.msg945361#new) Bob.

The link isn't working. Can you post the URL?
Title: Re: Early ON1 Photo RAW testing?
Post by: Cem on November 24, 2016, 09:18:31 am
The link isn't working. Can you post the URL?
Does this work? http://forum.luminous-landscape.com/index.php?topic=114584.msg945361#msg945361
Title: Re: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on November 24, 2016, 09:21:25 am
Yes, Thanks!

I wonder why that didn't show up in my search - maybe I searched the wrong forum area.
Title: Re: Early ON1 Photo RAW testing?
Post by: Bart_van_der_Wolf on November 24, 2016, 09:43:37 am
Yes, Thanks!

I wonder why that didn't show up in my search - maybe I searched the wrong forum area.

Hi Bob,

I'll be posting some of my findings (image crops) in this thread here when I get a bit more time. I'll re-convert with other software, in order to have the most up-to-date comparison (although the other software has not changed its algorithms, that I'm made aware of).

This will make it easier to track the progress if any, and after an official release on December 19th a new thread can be started to track the announced (sofar missing) features.

Cheers,
Bart
Title: Re: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on November 24, 2016, 10:05:54 am
I'll be posting some of my findings (image crops) in this thread here when I get a bit more time.

Excellent! I'll be looking forward to that.

Sadly, the initial results don't support my high expectations of a new, from the ground-up, RAW converter. Or reinforces the idea that such a thing is not a trivial endeavor.
Title: Re: Early ON1 Photo RAW testing?
Post by: Bart_van_der_Wolf on November 24, 2016, 02:48:53 pm
I'm busy with a technical analysis of the On1 Photo Raw conversion process but run into puzzling effects. The Raw conversion seems to have noise reduction and (over-)sharpening applied that cannot be influenced with user controls. The (over-)sharpening may be intentional for compensation of a relatively low-resolution conversion. It almost looks like there is a sharpening going on with different radii at different brightness levels.

It's a puzzling effect, but it does not produce artifact-free results, on the contrary.

To be continued, when I better understand what happens, and what to do about it.

Cheers,
Bart
Title: Re: Early ON1 Photo RAW testing?
Post by: Cem on November 24, 2016, 02:55:36 pm
This sounds serious Bart. Keep us updated please. :)

I'm busy with a technical analysis of the On1 Photo Raw conversion process but run into puzzling effects. The Raw conversion seems to have noise reduction and (over-)sharpening applied that cannot be influenced with user controls. The (over-)sharpening may be intentional for compensation of a relatively low-resolution conversion. It almost looks like there is a sharpening going on with different radii at different brightness levels.

It's a puzzling effect, but it does not produce artifact-free results, on the contrary.

To be continued, when I better understand what happens, and what to do about it.

Cheers,
Bart
Title: Re: Early ON1 Photo RAW testing?
Post by: Bart_van_der_Wolf on November 27, 2016, 07:37:25 am
It's a puzzling effect, but it does not produce artifact-free results, on the contrary.

To be continued, when I better understand what happens, and what to do about it.

I've made an animation of the cropped center of a single Raw file of my resolution target (http://www.openphotographyforums.com/forums/showthread.php?t=13217). The center crop has been zoomed in to 400% to make it easier to see what happens, so at 100% zoom or smaller the artifacts are less visible, but they do make postprocessing (e.g. sharpening) more error prone.

Some of the False color artifacts can be mitigated a bit by some of the conversters, but I used non-adjusted Raw conversions, with all sharpening and noise reduction disabled.

My personal favorite still is Capture One, which produces images with very high resolution with a minimum of artifacts that can cause issues in subsequent processing. I've shown a Version 8 conversion, but version 9 produces identical results. RawTherapee (not shown here) produces results that are very close to Capture One's.

Lightroom (version 5.7 produces indentical results to the current versions) is a bit behind in both resolution and false color artifacting. As said, the false color artifacts can be mitigated a bit, but there are other converters that have more effective tools for that. On occasion Lightroom will produce edge artifacts that are more prominent than those shown here.

OnOne Photo RAW is very aggressive in extracting detail, and that results in mazing artifacts that even limit resolution before it reaches the file limits. The artifacts and lower resolution produce a gritty look the closer one gets to the file's limiting resolution. In addition, the default conversion (without sharpening or noise reduction) already seems to apply significant sharpening that cannot be disabled or controlled, and that leads to halos on sharp edges. The shown region of the resolution target only has smooth sinusoidal features, so the halos are not that obvious although it tends to exaggerate the contrast of small detail.

As a bonus, I've added a crop from the Affinity Photo Beta version 1.5.0.39 for Windows. I'm not sure if the Mac version produces the same results. The Windows version is clearly a Beta version with some issues that need to be ironed out, but they are working hard to solve them before official release. The recent update, to my eyes, also improved/changed the Raw conversion quality a bit, so I'll reserve a final judgement for the final version. However, the current version is quite aggressive in extracting maximum detail, which results in some false color artifacting, and some mazing artifacts at the very limit of resolution. One could paint in some (Chroma) noise reduction to mitigate the most visible issues.

The image I used was shot on a Canon 1Ds Mark III, with the EF-135mm f/2.0 L lens at its optimal resolution aperture, on a heavy tripod. I made sure to get the best possible focus with that combination by bracketing the focus position with a focus rail, and picking the measured highest resolution frame. Raw conversions were all done without sharpening or noise reduction.

That also means that cameras without an anti-aliasing filter will cause even more prominent artifacts, and that also explains why Capture One does so well (given its legacy with having to deal with MF sensor images with large sensor elements without low-pass filter or microlenses).

The artifacts will probably not be as prominent with lower quality lenses or with slight defocus (even as it approaches the boundaries of the DOF zone). A bit of defocus blur, camera shake, diffraction, subject motion, will reduce resolution and thus make for a less critial Demosaicing and Raw conversion.

So my test will bring out the issues one can expect if everything else is optimal. With lesser technique the issues will be less obvious, but then the whole image will be of lower technical quality.

Cheers,
Bart

P.S. The green circle represents the Nyquist limit, so anything inside that circle is aliasing which one can only hope is not too obtrusive. As detail approach and exceed the Nyquist limit, the micro-contrast would ideally fade to uniform gray if we want to avoid all aliasing artifacts.
Title: Re: Early ON1 Photo RAW testing?
Post by: Jack Hogan on November 28, 2016, 04:15:58 am
Nice work, Bart.
Title: Re: Early ON1 Photo RAW testing?
Post by: Cem on November 28, 2016, 06:24:05 am
Thanks a lot for the hard work Bart, really appreciated. It confirms my "non-scientific" observations that the resolution is somewhat lower in the conversions done by the ON1 RAW compared to Capture One and LR. I think that ON1 will need to improve the raw conversion aspect of their software significantly if they want to attract users who want to extract the best possible results from their raw files.
Title: Re: Early ON1 Photo RAW testing?
Post by: kirkt on December 13, 2016, 10:04:56 am
Here's a short review from someone who has pointed out some of the things I, too, have noted in using the pre-release2 version of PhotoRaw.

https://youtu.be/Nlpb7hS-hCM

kirk
Title: Re: Early ON1 Photo RAW testing?
Post by: Alan Smallbone on December 13, 2016, 11:12:08 am
Kirk,

Thanks for posting that, interesting video. I too have some of the same concerns, but I think they will eventually get them worked out. Like anything new there are always growing pains. This is certainly and ambitious project that is quite complex, so I am willing to give them some time and will continue to try and play with it.

Alan
Title: Re: Early ON1 Photo RAW testing?
Post by: brandon on December 14, 2016, 04:25:53 am
Thanks Bart for a very useful (and telling) animated comparison with a high quality "target".
I have enjoyed ON1 for its filters and presets, and pre-ordered their RAW but will stay with LR or CO for raw and PS to call Topaz and other plugins for most images and ON1 10.5 for filter/presets avoiding their RAW for now at least.
Regards Brandon
Title: Re: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on December 20, 2016, 06:54:32 am
With the release version available yesterday, it will be interesting to see what the consensus opinion is on the raw conversion quality. I'm not a sophisticated tester, but my results suggest no difference between the pre-release and the release versions. ONONE doesn't produce the same level of detail as does Lightroom or Capture One - it's noticeably soft.
Title: Re: Early ON1 Photo RAW testing?
Post by: Cem on December 20, 2016, 06:57:58 am
The final release runs faster and more stable on my PC so far. That is definitely an improvement. The conversion quality isn't there yet. And it doesn't have a control for removing chromatic aberrations yet. Which should be an essential part of the raw converter imo.
Title: Re: Early ON1 Photo RAW testing?
Post by: Alan Smallbone on December 20, 2016, 04:21:01 pm
The final release runs faster and more stable on my PC so far. That is definitely an improvement. The conversion quality isn't there yet. And it doesn't have a control for removing chromatic aberrations yet. Which should be an essential part of the raw converter imo.

They have a list of planned additions and I am pretty sure it is on the list, there is a post on their site that lists the schedule, something like every 4-6 weeks there will be an update.

Alan
Title: Re: Early ON1 Photo RAW testing?
Post by: Cem on December 20, 2016, 04:34:31 pm
They have a list of planned additions and I am pretty sure it is on the list, there is a post on their site that lists the schedule, something like every 4-6 weeks there will be an update.

Alan
Yes, they mention it under Automatic Lens Corrections, which will be update #2 in 2017. I am going to keep on using the ON1 Photo Raw 2017 for the effects (filters) part, since it is many times faster than the previous (non RAW) version. I'll continue doing the raw conversion itself in either Capture One or LR. Just the speed improvement and the ability to switch individual filter layers on/off is enough improvement for me to justify the upgrade.
Title: Re: Early ON1 Photo RAW testing?
Post by: Shiever on December 20, 2016, 07:23:55 pm
I see that Bart has also included an early version of Affinity Photo.  Now that AP 1.5 is out and Luminar is out, how do they fit into the raw conversion "race".  I would like to get away from both LR and C1 with software that is "good enough" for x-trans RAW and appears to have a long term improvement program. 
Title: Re: Early ON1 Photo RAW testing?
Post by: Cem on December 21, 2016, 05:16:37 am
.... I am going to keep on using the ON1 Photo Raw 2017 for the effects (filters) part, since it is many times faster than the previous (non RAW) version. I'll continue doing the raw conversion itself in either Capture One or LR. Just the speed improvement and the ability to switch individual filter layers on/off is enough improvement for me to justify the upgrade.
Coming back to this, I have tried editing multiple photos in one go today. When I select multiple photos in LR and send them to ON1 Photo RAW for editing, only the first photo in the batch is updated by Photo RAW. The remaining photos in the selection are returned to LR without any updates. This process used to work with Photo 10.5, it doesn't work now. I have reported this to ON1 support, hopefully it will be fixed soon.
Title: Re: Early ON1 Photo RAW testing?
Post by: Bart_van_der_Wolf on December 21, 2016, 06:12:38 am
I see that Bart has also included an early version of Affinity Photo.  Now that AP 1.5 is out and Luminar is out, how do they fit into the raw conversion "race".  I would like to get away from both LR and C1 with software that is "good enough" for x-trans RAW and appears to have a long term improvement program.

Hi,

I have not specifically tested X-trans Raw conversion quality, but I've heard positive things about the Capture One conversions. How On1 Raw handles them is something that has to be verified but, given the issues with regular Bayer CFA demosaicing, it would surprise me a bit if it does better with X-trans files. Whether it's 'good enough', is up to the user.

Cheers,
Bart
Title: Re: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on December 23, 2016, 08:34:03 am
I've been attempting to compare the raw conversion of ON1 Photo RAW and Lightroom but am finding it maddening difficult to get both programs set to "default" for that comparison. Does anyone have any tips for getting both programs set to the "same" default adjustments?
 
Title: Re: Early ON1 Photo RAW testing?
Post by: Jack Hogan on December 23, 2016, 12:50:12 pm
I've been attempting to compare the raw conversion of ON1 Photo RAW and Lightroom but am finding it maddening difficult to get both programs set to "default" for that comparison. Does anyone have any tips for getting both programs set to the "same" default adjustments?

There is no 'standard' output.  Every raw converter gives its own version of what it considers to be a 'pleasing' rendition upon opening a file.  It's very subjective.  As far as ACR/LR is concerned I personally do not like 'Adobe Standard' for my camera, preferring 'Camera Standard' instead (which is Adobe's attempt to copy Nikon's 'Standard' Picture Control).

Jack
Title: Re: Early ON1 Photo RAW testing?
Post by: Shiever on December 23, 2016, 05:31:37 pm
On YouTube, the Modern Artist had a recent video where he is comparing various raw converters (LR,C1.ect) on low light, high iso images.  He noted that each converter had a different starting point and in his mind there is always some proprietary things happening with the algorithms. Further adjustments eventually gets you to a similar result.  I really do not know his level of knowledge/expertise, but his conclusions seem to match my experiences.  https://youtu.be/Y4WzaXrLwJI
Title: Re: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on December 23, 2016, 06:18:24 pm
The work he discussed on the video was focused almost exclusively on noise treatment for high ISO X-Trans files.

I have a hard time agreeing with him that they will all end up in a satisfactory place, given enough adjustments. The initial raw conversions counts - starting from a bad place is going to make it hard (if not impossible) to get to a good one. Hence my interest in which coverter provides a "better" starting point and if ON1 Photo RAW is already in the pack with the top converters, or if they have years of development to get where Adobe and Phase One are now.
Title: Re: Early ON1 Photo RAW testing?
Post by: john beardsworth on December 24, 2016, 04:22:06 am
What though is a "bad place"? Don't forget the human aspect here, because a flat or neutral default rendering (assuming default exists) can often give you a much better roadmap, revealing the image's final possibilities, than a more "pleasing" default may do.
Title: Re: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on December 24, 2016, 09:42:56 am
What though is a "bad place"?

I'm thinking in terms of image quality that can't be fixed later with a program's adjustment tools. All raw converters can't be equal, some must produce better demosaic results. I would think that the amount of detail extracted from the raw file would be a part of that. Perhaps there are others I'm overlooking?

The default starting point is less important to me because, as you point out, different starting points provide different base images to work from.
Title: Re: Early ON1 Photo RAW testing?
Post by: Jack Hogan on December 24, 2016, 12:33:28 pm
I'm thinking in terms of image quality that can't be fixed later with a program's adjustment tools. All raw converters can't be equal, some must produce better demosaic results. I would think that the amount of detail extracted from the raw file would be a part of that. Perhaps there are others I'm overlooking?

Raw converters do a number of menial tasks and a couple of critical ones, the rest is just pre-editing (you can read about a simplified breakdown of the key steps involved in rendering a raw file here (http://www.strollswithmydog.com/raw-file-conversion-steps/)).   What's critical today?

My opinion is that in late 2016 demosaicing is no longer critical: algorithms are tried, tested and stable - and they have been so for at least the last 5+ years or so, each with their known strengths and weaknesses.  Improvements have slowed to the point of being almost unnoticeable from year to year.  Same with sharpening and noise reduction.

Where I think they differentiate today is 1) a pleasing starting point upon first opening of the raw file, and 2) ease of squeezing camera DR into output device DR naturally.   Both of these features are very subjective but, subjectively, some do better than others.  They have mainly to do with getting color and (global, local, micro) contrast rendition in the desired ballpark out of the box.  Pretty well everything else can be done better and more easily in a good editor like PS.

Jack
Title: Re: Early ON1 Photo RAW testing?
Post by: scyth on December 24, 2016, 01:46:58 pm
What's critical today?
you totally forgot about the speed to reflect changes in controls in real time for example and that includes the art of GPU coding for the said algorithms, which few converters mastered
Title: Re: Early ON1 Photo RAW testing?
Post by: scyth on December 24, 2016, 01:49:09 pm
Pretty well everything else can be done better and more easily in a good editor like PS.
except that hoi polloi do not want to go to PS, hence you either aim for a niche or if you want to fight for marketplace (or what remains of it) then you have to do it all inside.
Title: Re: Early ON1 Photo RAW testing?
Post by: Jack Hogan on December 25, 2016, 05:16:09 am
Sure and sure, scyth. Plus for some people the darn DAM is a big deal, etc., etc.  I guess it depends whether one has a high volume or low volume flow.  If the latter, as far as image 'quality' is concerned, those are the two that stand out to me.  I sometimes use horrible and slow Capture NX-D because it renders the file well upon first opening - and  exactly as I had it set up in-camera; then do the rest elsewhere.  I know that that may not be everybody's cup of tea.

Jack
Title: Re: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on December 25, 2016, 10:20:03 am
My opinion is that in late 2016 demosaicing is no longer critical: algorithms are tried, tested and stable - and they have been so for at least the last 5+ years or so, each with their known strengths and weaknesses.

Does this suggest that ON1 has used existing demosaic algorithms, rather than develop their own? In would have thought the critical components of a demosaicing engine would have been patented by the likes of Adobe and Phase One. No?
Title: Re: Early ON1 Photo RAW testing?
Post by: Bart_van_der_Wolf on December 25, 2016, 01:07:14 pm
Does this suggest that ON1 has used existing demosaic algorithms, rather than develop their own?

Hi Bob,

My own analysis of the Raw conversions leads me to the conclusion that they used some sort of algorithm of their own that uses noise reduction and (too much) sharpening. Other converters do none of those, although they may apply a default to the subsequent controls that the user can then reduce to zero if they like.

Cheers,
Bart
Title: Re: Early ON1 Photo RAW testing?
Post by: Jack Hogan on December 26, 2016, 05:07:54 am
Does this suggest that ON1 has used existing demosaic algorithms, rather than develop their own? In would have thought the critical components of a demosaicing engine would have been patented by the likes of Adobe and Phase One. No?

What Bart said plus many demosaicing algorithms (including current champs) are in the public domain.  Folks like adobe simply customize them a bit for their workflow.  RawTherapee gives the option of trying a collection of several advanced ones, as used in well known raw converters.  Based admittedly solely on what I have seen in this thread, I think to understand ON1 raw conversion one can start by looking at their old filter suites.  While those may be fun in the creative department, they may appear a bit heavy handed when simply wishing to open a 'neutral' raw file.

Jack
Title: Re: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on December 26, 2016, 07:41:57 am
My own analysis of the Raw conversions leads me to the conclusion that they used some sort of algorithm of their own that uses noise reduction and (too much) sharpening. Other converters do none of those, although they may apply a default to the subsequent controls that the user can then reduce to zero if they like.

This is more in line with what I would have guessed: ON1 created their own algorithm, perhaps by combining and customizing publicly available ones.

What is interesting is that they have opted to "bake in" certain adjustments during the demosaicing that cannot be reversed using their adjustments UI. I would think that adjustment defaults would be added such that they could be adjusted by the user.

And it certainly makes comparing the base conversions between converters very difficult.

I suppose I'll come back and look at ON1 again, after they've added some of the more important (to me) features on their development roadmap. I give them kudos for having a more public development process.
Title: Re: Early ON1 Photo RAW testing?
Post by: dseelig on January 21, 2017, 03:47:49 pm
I tried today with the latest version but it crashes constantly so no go for me now
Title: Re: Early ON1 Photo RAW testing?
Post by: Alan Smallbone on January 25, 2017, 09:17:07 am
I tried today with the latest version but it crashes constantly so no go for me now

What kind of system? I have not had a crash yet with Win 10, 16gb ram and plenty of disk space.

Alan
Title: Re: Early ON1 Photo RAW testing?
Post by: mburke on January 26, 2017, 07:03:56 am
I have been using Photo Raw since the first version. Very early on I had 2-3 crashes. Haven't had one in a month. I've been very pleased with everything about PRaw. It should be even better as they add features. For the landscapes I do and the amount of photos I do it works really well. I use Win10, I-7, 16 g ram. They have some wonderful tutorials. It can be very powerful when using the protection and apply to functions on all of the effects.
Title: Re: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on January 26, 2017, 07:44:48 am
I have been using Photo Raw since the first version. Very early on I had 2-3 crashes. Haven't had one in a month. I've been very pleased with everything about PRaw. It should be even better as they add features. For the landscapes I do and the amount of photos I do it works really well. I use Win10, I-7, 16 g ram. They have some wonderful tutorials. It can be very powerful when using the protection and apply to functions on all of the effects.

Have you been satisfied with the RAW conversion quality compared to ACR (or Capture One)?
Title: Re: Early ON1 Photo RAW testing?
Post by: mburke on January 27, 2017, 06:54:21 am
Yes I have been satisfied. Early on the Highlights slider was a little weak but it seems to have improved. You can usually get to where you want. If you start with a somewhat well exsposed  pic you should be just fine. After you learn some of the protect sliders and apply to functions it is really a powerful piece of software. Good videos on their site. The more I use the more I like. Lots easier and cleaner for my use than the Lroom. I use an old copy of CS6 or Affinity for any things that ON1 can't do (pano, hdr, focus stack)
Title: Re: Early ON1 Photo RAW testing?
Post by: Bob Rockefeller on January 27, 2017, 07:27:15 am
Yes I have been satisfied. Early on the Highlights slider was a little weak but it seems to have improved. You can usually get to where you want. If you start with a somewhat well exsposed  pic you should be just fine. After you learn some of the protect sliders and apply to functions it is really a powerful piece of software. Good videos on their site. The more I use the more I like. Lots easier and cleaner for my use than the Lroom. I use an old copy of CS6 or Affinity for any things that ON1 can't do (pano, hdr, focus stack)

I'm thinking more about the actual, initial RAW conversion. How much detail is extracted? Are artifacts created? Remember all the fuss over the smearing of foliage by Lightroom in X-Trans files - that sort of thing.
Title: Re: Early ON1 Photo RAW testing?
Post by: mburke on January 27, 2017, 01:24:23 pm
I haven't noticed any real issues with artifacts. I also use an Epson 3880 and have done some very nice 13x19's. I use Qimage to print so I simply export a jpg or tiff to a print folder and let Qimage do its thing and then delete the file or not depending on usage. I find that using On1 Browse, Develop and Effects to be a very effective workflow. If I want to mess with changing sky or textures I can use their Layers module or else go to CS6 or Affinity. I use CS6 for panos and HDR. Over the next few months when ON1 has lens corrections and compare and a few other enhancements it should be a nice program.
Title: Re: Early ON1 Photo RAW testing?
Post by: Shiever on January 27, 2017, 03:27:27 pm
On1 knows that they have a problem with x-trans files and have announced that in their release in February there will be improved x-trans processing.  On1 also is promising to do x-trans compressed files in the Feb time frame.  Personally I have not seen the smearing effects on foliage.  I have seen other artifacts along edges (sawtooth effect).  There is also some CA in all of the processed files.  These are both especially obvious along tree lines where they meet the sky.  Hopefully, the Feb update will permanently be solved.