Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: PaulSchneider on June 09, 2015, 02:53:19 am
-
Hi guys!
Scroll down to raws, JPGs are worthless:
http://www.photographyblog.com/previews/canon_eos_5ds_sigma_24mm_35mm_50mm_photos/
Open on MAC with Iridient raw developer (free trial possible) and play around with the sharpening filters (reveal and hybrid).
To me, I was not excpecting this level of sharpness. Looks like a piece of MFD to me!
What do you think? In my view, Canon plus great glass ist not bad at all, it seems!
Colors are plasticky though!
Best
-P
-
Colors are plasticky though!
Hi,
The camera doesn't make plasticky colors, it just records them. Color rendering depends on the converter's algorithms, and the camera's color profile used ...
Cheers,
Bart
-
Be careful, the MF police will be here soon!
Unless you can convince them that the 5Ds is a MF camera, you are in deep trouble my friend! ;)
Cheers,
Bernard
-
Hi,
The camera doesn't make plasticky colors, it just records them. Color rendering depends on the converter's algorithms, and the camera's color profile used ...
Cheers,
Bart
Bart, coming from *you* of all people here, this statement really astonishes. You're certainly considerably more knowledgable than I am when it come to imaging. Maybe you need a couple of strong coffees and a helpful counterexample to jolt your brain awake? Well, the RED Dragon and its foobarred sensor provides it nicely. An accessory filter has been designed to improve the color performance of this sensor in ways which software cannot. And yes, texture loss may not necessarily be a hardware sensor issue proper, but it may be a side effect of the hardware choices which a manufacturer made for noise removal and signal processing.
And BTW, for once this post is not off-topic because clean color and texture really are one of the things which a lot of the old CCD backs were appreciated for.
http://www.newsshooter.com/2015/06/06/achtels-trueblue-optical-filter-for-red-dragon-and-red-epic-cameras-removing-unwanted-magenta/
(http://www.newsshooter.com/wp-content/uploads/2015/06/Screen-Shot-2015-06-06-at-9.49.44-AM-600x569.png)
-
Interesting to play with the downloadable 5Ds RAW files. BIG!!!
DxO Optics Pro 10 didn't have the modules to open them yet. RawTherapee will, and Corel just put out Patch 3 for PaintShop Pro X7 today that allows its RAW engine to open them.
However, the whites in the church ceiling (ISO 12,800) are really peppered with red/green noise in Corel PSP X7 III new RAW for them. I then opened it in Topaz Denoise 5 and had to go to RAW-Stronger to get rid of it and they do look better doing that. Hopefully DxO Optics 10 will get the module out to open them and correct the distortion along with their PRIME engine to see how it handles those files (Which may take several minutes based on their size and my knowledge of how slow the D800E files are processed.).
SG
-
I think a lot of people are going to be happy with this new camera.
-
My cat has just taken a photo with a new Samsung smartphone that, to my eye, looks like it was taken with an IQ380. Ok to post some pics here ?
-
The camera doesn't make plasticky colors, it just records them. Color rendering depends on the converter's algorithms, and the camera's color profile used ...
Bart, coming from *you* of all people here, this statement really astonishes.
Edmund, why? Do you have any evidence that the 5DS / 5DS R does produce plasticky colors?
Of course there are differences between cameras, due to the characteristics of the IR filtration and the chosen CFA, but I do not know of any credible evidence that the 5DS / 5DS R is poor in that respect. On the contrary, the Raw conversions in Capture One, and even in ACR, look pretty nice.
Well, the RED Dragon and its foobarred sensor provides it nicely. An accessory filter has been designed to improve the color performance of this sensor in ways which software cannot.
Maybe their choice of CFA made a correction filter more necessary. BTW, the marketing examples of the 'trueblue" filter effect are laughable, and their explanation is only partly correct. Since this is a thread about the Canon camera, I'm not going to waste my time in addressing the RED Dragon promotional claims.
And yes, texture loss may not necessarily be a hardware sensor issue proper, but it may be a side effect of the hardware choices which a manufacturer made for noise removal and signal processing.
Ah, so you do seem to understand what's going on. I also think the OPs comment was more related to noise reduced high ISO shots with a studio (low ISO optimized) camera. Software noise reduction may clobber detail if done wrong, but the camera does not write plasticky color data to the sensor.
Cheers,
Bart
-
Be careful, the MF police will be here soon!
Unless you can convince them that the 5Ds is a MF camera, you are in deep trouble my friend! ;)
Cheers,
Bernard
Yeah, you screwed up Paul.
Mind your back!
;D
Steve Hendrix
CI
-
Not to take any merits from MFD systems away! A lot of the value comes from other factors beside the pure raw file: back handling and integrated review tools, ipad integration, LS lenses, MFD booked and representation of space due to the bigger format, capture one integration with the whole system leading to great raw processing etc. What anyone is able or wants to pay for that is another topic.
But nonetheless, I was honestly surprised to see this quality from a canon sensor lens combo - the last camera I had from Canon was a 5d MK II and in conjunction with the then existing red-stripe lens offerings from Canon the files never exhibited this level of sharpness and detail.
... But please do not incarcerate me! Dear police officers! :-[ ;)
Yeah, you screwed up Paul.
Mind your back!
;D
Steve Hendrix
CI
-
Considering most of those shots were taken at less than optimum apertures, Capture One 8.3 makes these files sing and they look amazingly impressive. A quick play around with the files in C1 using the Phase One IQ350 profile gives very pleasing colours to my eye.
-
Hardware do have some effect on color, but no way near as much as the average photographer thinks. It's mainly a profile thing.
I think many photographers would gain from having a workflow which incorporated making own color profiles. You could then with much less effort, risk and angst move between different camera systems and raw converters. Manfacturers gain from maintaining the view that color is 90% about hardware and only 10% about profile, while it's rather the other way around.
-
Hardware do have some effect on color, but no way near as much as the average photographer thinks. It's mainly a profile thing.
I think many photographers would gain from having a workflow which incorporated making own color profiles. You could then with much less effort, risk and angst move between different camera systems and raw converters. Manfacturers gain from maintaining the view that color is 90% about hardware and only 10% about profile, while it's rather the other way around.
So you think the profiles made by the converter geeks are bad?
Edmund
-
So you think the profiles made by the converter geeks are bad?
Well, they are like printer profiles, or default screen settings. Made to impress the average user, generally not made to be accurate or, well, good. If you by converter geeks mean employees at say Adobe that make profiles for commercial products.
There are exceptions though, the MFD guys, C1 & Hassy have been very successful in designing successful looks desired by the pros. However it is still about a designed look, a heritage from the film days. I don't think digital should work this way. I think it's time for photographers to take command over their own color and develop workflows that is less dependent on the taste of raw converter and camera makers.
I haven't really investigated the issue but I'm assuming many smaller raw converters like iridient etc make neutral profiles rather than a look though.
Can photographers succeed equally well, or is it just too difficult so you must leave it to the manufacturers? I don't know for sure. I know I prefer using my own neutral profiles than using a readily packaged look, but on the other hand I do landscape photos not portraits. But I think a major aspect of the problem is that people don't try, the more people try the more and better techniques of how to work with your colors to get them the way you like without having to buy camera X Y or Z would then appear.
CCD believers are getting fewer day by day anyway, and color responses become more and more equal between cameras (that's why cross-using profiles can provide quite good results) so I expect the talk about camera-specific color will eventually fade.
-
Yes. I suggest this for anyone who wants to try:
http://www.xrite.com/online-color-test-challenge
Edmund
Well, they are like printer profiles, or default screen settings. Made to impress the average user, generally not made to be accurate or, well, good. If you by converter geeks mean employees at say Adobe that make profiles for commercial products.
There are exceptions though, the MFD guys, C1 & Hassy have been very successful in designing successful looks desired by the pros. However it is still about a designed look, a heritage from the film days. I don't think digital should work this way. I think it's time for photographers to take command over their own color and develop workflows that is less dependent on the taste of raw converter and camera makers.
Can photographers succeed equally well, or is it just too difficult so you must leave it to the manufacturers? I don't know for sure. I know I prefer using my own neutral profiles than using a readily packaged look, but on the other hand I do landscape photos not portraits. But I think a major aspect of the problem is that people don't try, the more people try the more and better techniques of how to work with your colors to get them the way you like without having to buy camera X Y or Z would then appear.
-
Yes. I suggest this for anyone who wants to try:
http://www.xrite.com/online-color-test-challenge
I don't think that expert skill in differentiating small differences in shades are necessary to make good color in post-processing. The adjustments are not on that detail level, if they were they would just break gradients.
It's helpful when selecting the correct white balance or doing various fine-tuning which photographers do all the time anyway. I don't think it makes sense romancing a default look when people are tweaking it like crazy anyway. Why not just start with a neutral rendering and then work from there.
That said I'm not claiming to have the absolute answer here. Maybe I shall take a look on those pictures to start with...
-
Very nice test! I got 4, rushed through the last one a little bit.
-
I don't think that expert skill in differentiating small differences in shades are necessary to make good color in post-processing. The adjustments are not on that detail level, if they were they would just break gradients.
It's helpful when selecting the correct white balance or doing various fine-tuning which photographers do all the time anyway. I don't think it makes sense romancing a default look when people are tweaking it like crazy anyway. Why not just start with a neutral rendering and then work from there.
That said I'm not claiming to have the absolute answer here. Maybe I shall take a look on those pictures to start with...
Actually yes, the adjustments are at that level. Go to a big fashion shop, and look at the lipstick and cosmetic shades. also, as a friend who does art repro told me 'if a painter has spent 15 years perfecting his blue, he wants to see his blue on the repro'. This is not photography, it is profile editing, a very specific task, for which many employ their own custom software, and first rate calibrated monitors, and good eyes.
Edmund
-
Actually yes, the adjustments are at that level. Go to a big fashion shop, and look at the lipstick and cosmetic shades. also, as a friend who does art repro told me 'if a painter has spent 15 years perfecting his blue, he wants to see his blue on the repro'. This is not photography, it is profile editing, a very specific task, for which many employ their own custom software, and first rate calibrated monitors, and good eyes.
That's repro. Making repro profiles requires a lot of knowledge and craftsman skill in measurement and target design, but you rely on instruments not eyes. Repro does not leave space for personal interpretation, it's about copying as accurately as possible.
You don't get to add things like making highlights warmer, shadows cooler and more saturated, and increasing separation of neutrals which are typical adjustments in the standard subjective profiles coming with the big commercial raw converters.
In repro your job is to copy. I'm no expert in repro but I've followed quite a few workflows and not a single one include the idea of hand-tuning colors to taste. You make a target, ideally with the same colorants as the artwork you're copying, you measure it with a spectrometer, and you shoot it and design a LUT profile that makes colors as accurate as possible.
Looking at typical fashion editing there's generally a lot of post-processing work, evening out colors, compensating for slight light differences, compensating for model not having full body makeup etc. To me it seems unlikely that these professionals actually gain from having a Phase One Image Professor tune a profile after his specific taste and light conditions. It would make much more sense to actual users if the standard in digital photography was to deliver as accurate colors as possible out of the box, rather than trying to differentiate cameras with "looks". If you make and sell cameras and also to some extent raw converters you gain from the widespread view that color is mostly a hardware thing and making good color is so difficult that it's better using a bundled generic profile made by experts with golden eyes rather than using your own eyes adjusting color for your specific scene and content.
-
Torger,
If you want a neutral profile, there's a very nice gadget from Image Engineering which has a bunch of narrowband filters and computes a profile for any illuminant with one click.
If you want to go and make edited profiles go ahead, no one is stopping you. If you live in America you are entirely free to have bad taste, it's called artistic innovation, and even if you live elsewhere, I don't think the cultural commissars will pursue you for color heresy. There are probably still some editors on the market, when I was doing this seriously I wrote my own editing suite, and so did the other color consultants I know who do camera color.
But if you continue to assert that camera color is just a software issue, then I would suggest you need to go out and shoot some more. You remind me of Thomas Knoll, a superb programmer and imaging expert who discovered Raw color quite late and decided to characterise cameras by shooting targets using an adjustable fluorescent lightbooth to select the color temperature. The result were those lovely Lightroom profiles, which characterized his lightbooth very well, and ensured that C1 stayed the top Raw converter for quite some time.
Edmund
That's repro. Making repro profiles requires a lot of knowledge and craftsman skill in measurement and target design, but you rely on instruments not eyes. Repro does not leave space for personal interpretation, it's about copying as accurately as possible.
You don't get to add things like making highlights warmer, shadows cooler and more saturated, and increasing separation of neutrals which are typical adjustments in the standard subjective profiles coming with the big commercial raw converters.
In repro your job is to copy. I'm no expert in repro but I've followed quite a few workflows and not a single one include the idea of hand-tuning colors to taste. You make a target, ideally with the same colorants as the artwork you're copying, you measure it with a spectrometer, and you shoot it and design a LUT profile that makes colors as accurate as possible.
Looking at typical fashion editing there's generally a lot of post-processing work, evening out colors, compensating for slight light differences, compensating for model not having full body makeup etc. To me it seems unlikely that these professionals actually gain from having a Phase One Image Professor tune a profile after his specific taste and light conditions. It would make much more sense to actual users if the standard in digital photography was to deliver as accurate colors as possible out of the box, rather than trying to differentiate cameras with "looks". If you make and sell cameras and also to some extent raw converters you gain from the widespread view that color is mostly a hardware thing and making good color is so difficult that it's better using a bundled generic profile made by experts with golden eyes rather than using your own eyes adjusting color for your specific scene and content.
-
I don't claim it being a software only issue, only much more a software issue than photographers in general tend to believe. The problem as I see it is that photographers ditch cameras based on color rendition without first ever trying making their own profile. It's hard to blame them though, the availability of easy to use and still good profiling software is limited.
I have indeed my own software, http://www.ludd.ltu.se/~torger/dcamprof.html, but I'm not claiming it's easy to use. I'm well aware of Image Engineering's products, as my own profiling software is capable of processing SSFs. SSF-based profiles targeted for specific spectral data of your particular interest are very nice indeed, but it requires expensive gear to measure (or ebay stuff and DIY electronics skill). Traditional target profiles do quite well too.
I doubt Lightroom's profiles are currently designed that way, many (most?) of the ones I've seen apply looks and are not that accurate, and the type of accuracy errors is not related to using a spiky fluorescent source as calibration illuminant (which indeed is not a good idea) but are rather "look" tunings. I haven't seen all their profiles though and I have noted design has changed over the years.
Anyway, I think it's wrong to draw conclusions on a camera's color performance based on default rendering in some raw converter, and that's exactly what photographers do all the time.
Obviously profiling is a research area for me though, lots of things still to discover so I may change my views with more knowledge. Some aspects of photography reminds me of my HiFi audio days though, were people could absolutely hear a difference of the direction of cables to speakers (except in blind testing of course but that does not count in their world). I think I see a few of those things in photography too, but one could always say my eyes are bad... :)
-
Torger,
Yes, people buy cameras or even whole systems eg. Phase in order to get a baked-in look.
Would most of them be capable of recreating something useful themselves using editing software? I'm not so sure.
This debate is even stronger in the video domain, where Raw data is usually not available.
Edmund
I don't claim it being a software only issue, only much more a software issue than photographers in general tend to believe. The problem as I see it is that photographers ditch cameras based on color rendition without first ever trying making their own profile. It's hard to blame them though, the availability of easy to use and still good profiling software is limited.
I have indeed my own software, http://www.ludd.ltu.se/~torger/dcamprof.html, but I'm not claiming it's easy to use. I'm well aware of Image Engineering's products, as my own profiling software is capable of processing SSFs. SSF-based profiles targeted for specific spectral data of your particular interest are very nice indeed, but it requires expensive gear to measure (or ebay stuff and DIY electronics skill). Traditional target profiles do quite well too.
I doubt Lightroom's profiles are currently designed that way, many (most?) of the ones I've seen apply looks and are not that accurate, and the type of accuracy errors is not related to using a spiky fluorescent source as calibration illuminant (which indeed is not a good idea) but are rather "look" tunings. I haven't seen all their profiles though and I have noted design has changed over the years.
Anyway, I think it's wrong to draw conclusions on a camera's color performance based on default rendering in some raw converter, and that's exactly what photographers do all the time.
Obviously profiling is a research area for me though, lots of things still to discover so I may change my views with more knowledge. Some aspects of photography reminds me of my HiFi audio days though, were people could absolutely hear a difference of the direction of cables to speakers (except in blind testing of course but that does not count in their world). I think I see a few of those things in photography too, but one could always say my eyes are bad... :)
-
Yes, people buy cameras or even whole systems eg. Phase in order to get a baked-in look.
Would most of them be capable of recreating something useful themselves using editing software? I'm not so sure.
This debate is even stronger in the video domain, where Raw data is usually not available.
I'm aware that I might do wishful thinking. I need to get much deeper into profiling and camera color than I currently are to drop my humbleness.
What I can say from a scientific standpoint is that current modern cameras are more similar between models than they used to be. For a fixed high CRI light, a fixed palette and making a profile from a purpose-made target (ie repro use case) cameras of different brands will produce virtually identical output, if not one's doing something wrong.
However when a camera is facing low CRI lights or changing light conditions the differences in hardware start to show, cameras with tuned profiles to match a look under a specific light condition will start to deviate under a different one. This is an interesting aspect whose practical effects I don't know that much about. It's possible some cameras work significantly better in varying light conditions than others.
I don't reject the idea that maybe it's so difficult to create a pleasing look that many need to rely on bundled profiles. I'd like to investigate further. However I'm very skeptical about that any significant trace of that pleasing look is left when typical professional post-processing work has been applied, and I'm very skeptical about that those that have do that type of post-processing actually need some specific camera to get better color, they should do good with any reasonable camera. It's extremely difficult to see which camera is which if we get post-processed images in our hands, unless you look for lens artifacts and similar and draw conclusions from that.
-
I don't claim it being a software only issue, only much more a software issue than photographers in general tend to believe. The problem as I see it is that photographers ditch cameras based on color rendition without first ever trying making their own profile. It's hard to blame them though, the availability of easy to use and still good profiling software is limited.
Indeed.
Anyway, I think it's wrong to draw conclusions on a camera's color performance based on default rendering in some raw converter, and that's exactly what photographers do all the time.
That's also what I've been saying, the camera is not at fault, it's the rendering to color that needs to be improved (before blaming the wrong party).
Cheers,
Bart
-
Indeed.
That's also what I've been saying, the camera is not at fault, it's the rendering to color that needs to be improved (before blaming the wrong party).
Cheers,
Bart
All cameras will fail the luther ives condition differently, and thus render in differently strange ways. Also, the profile has culturally adjusted skin tones baked in ...remember, there used to be orange, yellow and green boxes of film
-
As someone who has made the effort to make profiles for every camera he has owned in recent times, I can safely say that the hardware choices are not completely overcome by profiling.
-
All cameras will fail the luther ives condition differently, and thus render in differently strange ways. Also, the profile has culturally adjusted skin tones baked in ...remember, there used to be orange, yellow and green boxes of film
Was there ever a more persuasive argument in favour of the Leica Monochrom ? ;D
M
-
Was there ever a more persuasive argument in favour of the Leica Monochrom ? ;D
M
Manoli,
You have spoken masterfully like a true Sophist. ;D
-
Was there ever a more persuasive argument in favour of the Leica Monochrom ? ;D
It fails the Luther Ives condition even more than most Bayer CFA filtered sensors. It also has one of the worst color reproduction characteristics ;)
All cameras will fail the luther ives condition differently, and thus render in differently strange ways.
The differences between the CFAs are relatively small if compared to human vision (which is significantly different), so they will all be similarly wrong.
Also, the profile has culturally adjusted skin tones baked in ...remember, there used to be orange, yellow and green boxes of film
There are many kinds of profiles possible. Canon also delivers their cameras with a variety of choices (e.g. Neutral, Faithful, Daylight, etc.) out of the box, and additionally offers a Picture Style Editor which allow to create one's own 'style' to be used by the camera and/or their Raw converter. Raw converters like Capture One allow to create and use user edited (ICC) profiles, and ACR allows to use tweaked DNG profiles. So there is not a "the profile", there is a vast number of choices, and options for users to create their own.
All that demonstrates that it's not meaningful to blame a camera in isolation, there are too many other factors that also have a much more significant influence on the resulting color rendering.
Given that the 5DS / 5DS R is not designed as a high ISO camera also means that more attention could be given to the choice of CFA filter properties, instead of being more transparent and less selective to gain some speed benefits.
Cheers,
Bart
-
My cat has just taken a photo with a new Samsung smartphone that, to my eye, looks like it was taken with an IQ380. Ok to post some pics here ?
Love it...
Maybe we can start a Canon page...oh wait, they have their own groups. ;D