Luminous Landscape Forum

Raw & Post Processing, Printing => Digital Black & White => Topic started by: Mousecop on April 14, 2017, 11:55:01 AM

Title: Monochrome Camera (conversion) vs Software Conversions?
Post by: Mousecop on April 14, 2017, 11:55:01 AM
Hi folks, sorry if this has come up before. I'm wondering if anyone has some experience comparing dedicated monochrome hardware to converting to B&W in post.

I'm fairly comfortable doing B&W conversions in LR, and the ability to tune specific channels gives me a lot of control. At the same time, hardware B&W reputedly produces better quality images in general.

If it matters, I'm limited to sensor conversions. Leica Monochrom is a bit much for me.

Assuming sensor sizes are the same, has anyone found that hardware B&W produces significantly better output?
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: unesco on April 14, 2017, 03:30:54 PM
Assuming sensor sizes are the same, has anyone found that hardware B&W produces significantly better output?
yes, at least my a few tries of Leica Monochrom gives more details, better tonality, better rendering in shadows includingh microcontrast (and higher ISO) - previous version of Leica compared to my EOS 6D.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Guillermo Luijk on April 14, 2017, 06:09:19 PM
Hardware (only) conversion doesn't exist. All B&W conversions, either in the camera or in your laptop, are ultimately performed by some piece of software. Said that, there is no reason why B&W conversions in your computer couldn't be the same or better quality than in camera conversions since you can have more processing power and there is no such speed requisites as in building camera JPEG's. Another story is that some camera produces such beautiful B&W images that you don't manage to match them on your computer.

Regards
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Eric Myrvaagnes on April 14, 2017, 07:22:09 PM
As I understand it, the Leica Monochrome's big advantage is the fact that every pixel records luminance only, no R G and B.

I have a friend who has done superb B&W work for many years with view cameras and film. He now scans his negatives and prints them digitally. He actually now also owns and uses to good effect a Leica Monochrome, and he does beautiful work with it.

As for me, I wouldn't want one even if I could get a new one for under a hundred dollars, because I would lose all the control I have in LightRoom or PhotoShop to decide what colors in the scene to make lighter and which darker.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: rdonson on April 14, 2017, 11:28:05 PM
As I understand it the Leica Monochrom is truly a monochrome sensor as Eric describes.

I love B&W.  One of the advantages of my Fuji X-T1 and X-T2 is the ability to shoot RAW+JPG while seeing the Fuji B&W film simulations in the EVF and LCD.  If I used a Leica Monochrome I'd think I'd have to go back to using Yellow, Red and Green filters in front of the lens.  No thanks.  Too many advantages for me sticking with an RGB sensor.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: BartvanderWolf on April 15, 2017, 01:16:15 PM
Assuming sensor sizes are the same, has anyone found that hardware B&W produces significantly better output?

Hi,

From a tonality point of view, no a monochrome sensor does not produce better output, not even insignificantly.

When you look at the spectral sensitivity curve of a regular sensor as is used for photography, see fig. III.11 (http://light-measurement.com/spectral-sensitivity-of-detector/) at the left side of the spectrum, the response is low in the blue end of the visible spectrum, and it increases towards the red end of the spectrum. This doesn't even resemble a panchormatic film's response.

So without the use of an additional (bluish-cyan) lens filter, the tonal reproduction will be relatively red oriented (darker blues, pale reds), like shooting through a light orange filter.

The benefit of shooting with a Bayer CFA filtered sensor is that you'll get huge amounts of control over how the final tonality will be done, based on color differences. The only 'cost' is a slightly reduced sensitivity compared to an unfiltered monochrome sensor. Tools like Topaz Labs B&W Effects (http://www.topazlabs.com/bweffects), or the now free Google Nik Silver Efex Pro (https://www.google.com/nikcollection/products/silver-efex-pro/), or the OnOne products and others, offer superior control and results.

So software conversion is, IMHO, much more flexible and can deliver better results.

Cheers,
Bart
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Ferp on April 15, 2017, 06:42:42 PM
Hardware (only) conversion doesn't exist.

Mousecop's request was for "hardware B&W".  In addition to the Leica Monochrom, there are niche services that will turn a color digital into B&W only by removing the Bayer sensor.  I'd assume that those options were what he was referring to, although it wasn't clear, as setting a Fuji to one of the B&W simulations and shooting in JPG could also be described as hardware B&W.  It seems to me that you've really got to want a modified camera to buy one or have one converted.  Personally I couldn't justify the cost of a Leica, not just because it's a lot of money but also because I doubt that the improvement would be worth the additional cost (YMMV).  And because like Bart I appreciate the flexibility that converting to B&W from color provides.  For me it's one of the advantages of the digital age.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: unesco on April 16, 2017, 05:49:24 PM
1. A discussion about software vs hardware BW conversion does not make sense. All hardware solutions are in fact software based this way or another and all software is run on hardware. I used to design both.

2. RGB conversion has its advantages - flexibility in manipulation and many other ones. On the other hand Leica Monochrome gives such a beautiful rendering of details, especially in dark areas plus wonderful tonality incomparably better than my EOS 6D after any type of BW conversion I have tried. If I could afford the overall L M ecosystem, I would definitely grab one.

3. Quite often I wonder why Canon or Nikon don't produce BW version of their FF cameras. I suppose there would be much more customers for them than for astro modified models.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Ferp on April 16, 2017, 06:35:00 PM
1. A discussion about software vs hardware BW conversion does not make sense. All hardware solutions are in fact software based this way or another and all software is run on hardware. I used to design both.

Once you use the term "BW conversion" then I guess that's right.  But the Monochrom and any cameras modified to remove the Bayer layer don't capture any color information, as I understand it, and so I don't see how you could describe those hardware solutions as being software-based B&W (obviously the camera still has software). 

The OP asked about "comparing dedicated monochrome hardware to converting to B&W in post".  That to me implied comparing a camera that either by design or modification doesn't capture color information, with converting from color to B&W, either in-camera or out-of-camera.  The heading for this thread was " Monochrome Camera (conversion) vs Software Conversions?", and I assume that this was short-hand for a camera that doesn't capture color information, either by design or modification (physical conversion of the camera).

Your observations in #2 about the Monochrom address the question as I understand it, although personally I would need to see this myself in order to be convinced.  What remains unanswered is whether those observations also apply to a modified camera.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Guillermo Luijk on April 16, 2017, 08:58:23 PM
The OP asked about "comparing dedicated monochrome hardware to converting to B&W in post".

The OP titled his post as "Monochrome Camera (conversion) vs Software Conversions". If that is not the same thing as software based in-camera B&W conversion then I'm a shaolin monk. I think he really made several questions in a bit confusing way.

BTW I proposed here:

http://forum.luminous-landscape.com/index.php?topic=117366.0

A way to obtain B&W from an RGB sensor maximizing SNR.

Regards!
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: GrahamBy on April 17, 2017, 12:47:59 PM
The OP titled his post as "Monochrome Camera (conversion) vs Software Conversions". If that is not the same thing as software based in-camera B&W conversion then I'm a shaolin monk.


Get the robes out. If you take the Bayer filter off the sensor, no colour info is ever captured: the spectral sensitivity is determined purely by the sensor. That sounds pretty hardware to me.

I still don't think it's a good idea: at best you win a stop of sensitivity... I just printed out a portrait made under available light at a party, shot at 12800iso. It has way less grain than anything you could do with 35mm film at any iso. For people, I would want more resolution. For landscapes, maybe you really do want that extra-super-duper high quality, but then you can get out a tripod.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Guillermo Luijk on April 17, 2017, 01:18:55 PM


If you take the Bayer filter off the sensor, no colour info is ever captured

So there is no conversion either. Let the OP clarify what he meant rather than doing your own assumptions.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: joofa on April 17, 2017, 03:18:30 PM
The benefit of shooting with a Bayer CFA filtered sensor is that you'll get huge amounts of control over how the final tonality will be done, based on color differences. The only 'cost' is a slightly reduced sensitivity compared to an unfiltered monochrome sensor.

Not sure if the 'cost' is a 'slightly reduced sensitivity'. Please see the image below that shows two actual images acquired with identical optical parameters with and without Bayer CFA with the same (type of) sensor. Please notice the means on the right side.

(http://djjoofa.com/data/images/bw_col_both.jpg)
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: GrahamBy on April 17, 2017, 03:32:38 PM

So there is no conversion either. Let the OP clarify what he meant rather than doing your own assumptions.

You start with an RGB camera. You convert it to a monochrome camera by removing the filter.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Guillermo Luijk on April 17, 2017, 05:34:10 PM
Not sure if the 'cost' is a 'slightly reduced sensitivity'. Please see the image below that shows two actual images acquired with identical optical parameters with and without Bayer CFA with the same (type of) sensor. Please notice the means on the right side.

(http://djjoofa.com/data/images/bw_col_both.jpg)
Your sample clearly shows the increased sensitivity derived from filter elimination, but reminds me of other benefits of keeping the filter array: increased sensor progressiveness, i.e. dynamic range.

I criticised the Leica Monochrome for paradoxically having a more 'digital' response than RGB cameras, since RAW channel individual exposures act as a dynamic range enhancer on RGB sensors.

Normally the B/R channel clip around 1,5 stops later than the G channel. While in a Leica Monochrome you have to be very careful about clipping the highlights, on RGB sensors you have around 1,5 extra stops of highlight headroom since B&W (actually also colour information) can be reconstructed from a single RAW channel.

The sun in this scene:

(http://www.guillermoluijk.com/misc/sun.jpg)

Has the following RAW data in EV along the red line:

(http://www.guillermoluijk.com/misc/suncanalesraw.png)

A monochrome highlight reconstruction strategy (dcraw -H 2):
(http://www.guillermoluijk.com/misc/suncanalesreveladoraw.png)

Inpaint colour algorithm (RawTherapee's colour propagation):
(http://www.guillermoluijk.com/misc/suncanalesreveladorawrt.png)

Playing ETTR on a monochrome sensor without accurate RAW histograms is more dangerous than doing it on RGB sensors. Moreover, saving highlights information on a monochrome sensor yields to overall severe underexposure while a RGB sensor still provides a good SNR on the G channel to save the shadow areas. RGB colour filter array acts as an HDR-like sensor: B/R save the highlights, G saves the shadows.

Regards.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Ferp on April 17, 2017, 07:40:19 PM
I pity the poor OP Mousecop.  Ask a simple (but slightly ambiguous) question and the thread wanders far away.  Guillermo first tries to deny that its about hardware modification vs Monochrom and then writes a thesis about why it's a bad idea.  In such situations the OP is never heard from again.  I suspect that all he wanted was some first-hand experience, but I think he's out of luck in this forum.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Guillermo Luijk on April 17, 2017, 11:01:27 PM
Look for a life worth being lived Ferp.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: joofa on April 18, 2017, 11:05:13 AM
Your sample clearly shows the increased sensitivity derived from filter elimination, but reminds me of other benefits of keeping the filter array: increased sensor progressiveness, i.e. dynamic range.

I criticised the Leica Monochrome for paradoxically having a more 'digital' response than RGB cameras, since RAW channel individual exposures act as a dynamic range enhancer on RGB sensors.

Normally the B/R channel clip around 1,5 stops later than the G channel.

Playing ETTR on a monochrome sensor without accurate RAW histograms is more dangerous than doing it on RGB sensors. Moreover, saving highlights information on a monochrome sensor yields to overall severe underexposure while a RGB sensor still provides a good SNR on the G channel to save the shadow areas. RGB colour filter array acts as an HDR-like sensor: B/R save the highlights, G saves the shadows.


I think you are right about some DR advantages. Monochrome and color cameras have different uses. So it is not always a question which is universally better.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: GrahamBy on April 18, 2017, 12:22:39 PM
since RAW channel individual exposures act as a dynamic range enhancer on RGB sensors.

Excellent point. Of course in utopia, we could imagine replacing the RGB Bayer filter with a grey-scale Bayer filter, where all cells would be in their linear range for most of the exposure range, but with different sub-sets remaining operative at the extremes : a sort of built in HDR facility.

In fact I believe it has been done: someone somewhere was recalling an early digital sensor supposed to help wedding photographers, by reserving a few lower-sensitivity cells to capture highlights from eg wedding dresses. In an ancient technology where the basic cell DR was far less than today.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Manoli on April 18, 2017, 12:27:25 PM
A Fetishist’s Guide to the Monochrom (Part 3) (http://www.ultrasomething.com/2012/12/a-fetishists-guide-to-the-monochrom-part3/) : pretty much explaining Guillermo's post in layman terms (sort of ..)
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Telecaster on April 26, 2017, 01:17:50 AM
When I use b&w film I rarely shoot "naked" (that is, sans filter).  :)  With a monochrom(e) camera it would likely be the same. Yet I often use an orange filter, which gives a visual result not far off what I'd get from a Monochrom without any filter. Hmmm…

Anyway photosite counts are at the point where you can sacrifice some spatial resolution during RAW conversion and get higher tonal res in return, and on an image-by-image basis. One RGBG photosite square equaling one output pixel or a similar scheme. It'd be nice if software creators took closer note of this (I say, for approx. the 100 millionth time since ~2003). [Edit begin]The point of this being: you could use the RGBG matrix for each output pixel as a set of filters for monochrome output. Way more flexible than a set of fixed-value filters screwed onto a lens mounted on a mono camera. In RAW conversion you could also skip the demosaic-to-RGB-color stage and go directly from Bayer data to (filtered) monochrome.[Edit end]

-Dave-
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Mousecop on May 09, 2017, 11:50:46 AM
Sorry for the delay, folks. For my own purposes, I've concluded I prefer the greater control in post. Anyway....


1. A discussion about software vs hardware BW conversion does not make sense. All hardware solutions are in fact software based this way or another and all software is run on hardware. I used to design both.
Hardware = monochrome sensor = sensor without a colored array (Bayer, X-Trans etc)

Software = starting with a color image, and using tools in post to convert to B&W (LR, Silver Efex etc)

That should've been obvious from the post, where I was comparing sensor conversions and the Leica Monochrom to LR  ;)


Quote
2. RGB conversion has its advantages - flexibility in manipulation and many other ones. On the other hand Leica Monochrome gives such a beautiful rendering of details, especially in dark areas plus wonderful tonality incomparably better than my EOS 6D after any type of BW conversion I have tried.
Yes, I'm asking if the overall image quality of a monochrome sensor is superior enough to offset the loss of control in post.

I suppose I should have requested some side-by-side examples. ;)
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Mousecop on May 09, 2017, 11:52:01 AM
I pity the poor OP Mousecop.  Ask a simple (but slightly ambiguous) question and the thread wanders far away.  Guillermo first tries to deny that its about hardware modification vs Monochrom and then writes a thesis about why it's a bad idea.  In such situations the OP is never heard from again.  I suspect that all he wanted was some first-hand experience, but I think he's out of luck in this forum.
Not really, I just made up my mind, and got distracted from the discussion by actually taking photos ;D
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: StoryinPictures on June 18, 2017, 05:34:47 PM
I'm surprised no one has brought up the Sigma cameras, which are fully capable of being used as a Leica Monochrome, using the data from the top "blue" layer. It is also possible to control how they are rendering color by taking advantage of the information in the lower layers.

The new Sigma SD Quattro and Quattro H are affordable alternatives to the Leica M. The non-H version can be had in a kit with a 30mm f1.4 lens for $1000 US.

The firmware updates mean all the current Quattros can also be set up to output DNG files, so they can be processed using Capture One, Camera RAW or LR.

That should give us enough fodder for a couple pages... :)
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: jevidon on June 18, 2017, 07:44:54 PM
This may have already been said; there being way too much conversation in this thread, but I believe that the Leica MM has neither a Bayer or antialiasing filter and therefore the sensor receives the image information without being filtered and therefore produces a higher quality image than is possible in a conventional digital camera whether equipped with a Bayer, Foveon or Fuji XTrans filter.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: StoryinPictures on June 24, 2017, 02:57:48 AM
This may have already been said; there being way too much conversation in this thread, but I believe that the Leica MM has neither a Bayer or antialiasing filter and therefore the sensor receives the image information without being filtered and therefore produces a higher quality image than is possible in a conventional digital camera whether equipped with a Bayer, Foveon or Fuji XTrans filter.

Mostly correct, but not in the case of Foveon. No AA filter and direct read from every pixel point if you take data from just the top layer.

This makes it on par with the Leica MM in both regards.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: Guillermo Luijk on June 24, 2017, 05:03:45 AM
I believe that the Leica MM has neither a Bayer or antialiasing filter and therefore the sensor receives the image information without being filtered and therefore produces a higher quality image than is possible in a conventional digital camera whether equipped with a Bayer, Foveon or Fuji XTrans filter.

I wouldn't dare to say supressing the AA filter produces higher quality images. An aliased image has never been a good quality image, and aliasing can hardly be eliminated in postpro since it means the valid data has been corrupted by high frequency interference.

Regards

Enviado desde mi ALE-L21 mediante Tapatalk

Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: jisner on July 16, 2017, 10:39:38 PM
I wouldn't want one even if I could get a new one for under a hundred dollars, because I would lose all the control I have in LightRoom or PhotoShop to decide what colors in the scene to make lighter and which darker.

Agree!  It's much easier to adjust tonality in color before doing a B&W conversion as the last step.  You can use color ranges and color channels to make targeted tonal adjustments.
Title: Re: Monochrome Camera (conversion) vs Software Conversions?
Post by: TommyWeir on July 17, 2017, 08:26:26 AM
The one thing which may differentiate the in-camera conversions are particular processes owned by the camera manufacturers, I understand that once you choose it in your RAW+JPEG settings, Fuji will bake in their own take on the Acros grain for example in their latest series of kit.  Given Fuji's history, their understanding of grain or how highlights and shadows fall off with various emulations of their stock has very pleasing results if emulating film stock is your aim...

The conversion tools within software applications may well be less sophisticated despite the power of the machines running them.   I see noticeable differences between the in camera JPEGs and any efforts attempting their recreation in C1 or Lightroom.