Chuck Coyne wrote:
I agree the CCD Fever and Monochrome CCD are very good but I also have some fun with the Fuji Film and Kodal Kodachrome as well
Interesting. The Kodachrome emulation is the only Cobalt purchase that has disappointed me. My own fault since I realised too late that what I really want is an emulation based not on the work of Ernst Haas or on Steve McCurry’s last roll but rather on that of Fred Herzog.
I'm sorry, but here I am again with some feedback.
I was noticing that the new Canon R5 emulation is still giving way too much saturation and what's even more annoying is that the greens get a different hue. They look almost nuclear.
Here's the DPR test scene that easily shows the proof of what I'm saying.
EOS R5, converted with DPP, standard profile:
A9, Cobalt EOS R standard emulation:
Again, this isn't even close. Look at that overal big difference in saturation + the greens look completely off. Just compare the green 'vegetation' in the bottom left corner.
Here's another example, this time of the 5DIV profile:
5DIV, converted in DPP, standard profile:
A9, Cobalt 5DIV standard profile:
The greens in this one are much much better than in the flawed R5 profile, but still... look how diffent some color hues are + the saturation again.
(All examples are white balanced after changing the profile)
I could go on and on showing these very obvious differences.
The claims I've read in this thread about how close the digital emulations are just don't hold true (as I've been showing before here).
Jochenb,
I was reading that and I said to myself "how this could be, I personally carefully checked all the pipeline and the result using the only proper way".
After that I found why you're missing the point.
You are comparing different cameras with different lenses, and is still something who is inside the "averaging" with our profiles cause the variability from lenses is not so big on modern lenses.... Not only that, you are comparing different raw editors for different brands.. so Sony on Adobe Vs Canon on dpp. Last but not least you are using different pictures... Yes, dpreview is not even far from a benchmark for colours and exposure and details as they clam, they are not coherent between cameras for basic settings and that pictures are just a test not made for colours and calibration.
If you want to test it, you should do at least a test comparing the R5 on dpp and another camera on Adobe ( knowing there is a difference in raw develop on the engine about the curve) for the same picture or the 5dmkiv emulation using the legacy as reference and the same picture on Lightroom for both cameras. And for same picture I mean something coherent (same moment, same light, same focal) or even the same file using the same camera would be the ideal test. I can't upload right now references but I already did same pages back and my results are not so different because I lowered at the minimum the variables.
And in the past I did a wedding mixing 5d4, a9 and A7r3... Using the Cobalt 5d4 for the final pictures, so I am the first user and tester for that.
Please try again using the same logic , perhaps taking some charts from imaging resource and reducing the variables when possible.
No I really can't agree with that at all Ulysseita.
I use these DPR scenes now just because it's an easy way to proof what I'm seeing in my own test shots between Canon and Sony cameras.
A different lens doesn't cause such big hue shifts and saturation difference at all. For example those greens never go that nuclear looking on the actual Canon files. Yet they do when I apply some Cobalt emulations on my Sony files.
You also talk as if such examples are shot in COMPLETELY different circumstances, light,... while they are clearly not.
The reason why I'm so vocal about it is because you literally say this on your website:
"We made a perfect emulation of the output coming both from Canon Digital Photo Professional software and the Jpegs SOOC."
So how can you say I'm wrong? I keep seeing excuses about it being user error.
I'm literally using your emulations made for Lightroom/Adobe Camera RAW and compare them to the conversion of Canon DPP. Your website claims such emulation is perfect, but they aren't even close to a perfect match at the moment.
Before I would be able to offer other samples(I am afk) please use Imaging resource images or, do what we did comparing real world shots of different cameras on same scene at the same time, the only proper way to start this topics.
I can't give accountability to the dpreview samples, if you have real shots between Canon and Sony of the same subject on same light, please show us, I will be happy to talk about that.
Imaging Resource doesn't have RAW files of their test scene for the A9.
Specifically showing the issues with the greens again:
This is grass how it looks on the R5, converted with DPP standard profile:
This is how grass often looks on the A9 with the Cobalt R5 standard profile:
It's easy to notice how the Cobalt profile shifts those greens to an almost "nuclear", unnatural looking grass.
Completely confirming what I already showed with the DPR test scene.
BTW, this is how it looks with the Cobalt neutral profile (the base one, not an emulation):
No issues here and similar to the Canon DPP results. Once again confirming how off the Canon emulation is.
We can see a just different reading from the different sensors of the scene for resolution, contrast and shadows and highlights, subtle differences, but the color correction has been done.
PS, off course, if on comparable samples like this, someone will find problems, we are always here to improve it!
Jochenb wrote:
Imaging Resource doesn't have RAW files of their test scene for the A9...It's easy to notice how the Cobalt profile shifts those greens to an almost "nuclear", unnatural looking grass. Completely confirming what I already showed with the DPR test scene
Ulysseita wrote:
Before I would be able to offer other samples…please use Imaging resource images or, do what we did comparing real world shots of different cameras on same scene at the same time, the only proper way...I can't give accountability to the dpreview samples, if you have real shots between Canon and Sony of the same subject on same light, please show us, I will be happy to talk about that.
@Jochenb - Do you think that you could turn the emotion down, say 10 notches? I know English is not your native language, but you're coming across as screaming — and, as @Ulysseita states, you're not comparing like with like. I am writing this because I feel that this thread is important for many photographers because, as I see it, Cobalt-Image offers the most accurate profiles that can be purchased (as opposed to making one's own) — and the simulation presets are of great general interest. Of course, you should provide your comments and suggestions, which can be helpful for a software developer, but, in the context of this thread, my suggestion would be for you to take this particular discussion to PMs and, then, only post here what is comparable or the conclusions.
@Fred Miranda - I'm grateful to you for starting this valuable thread. You may want to consider deleting the last few exchanges that I've referred to for the sake of clarity, so that this thread can remain useful for photographers who are interested in the Cobalt-Image profiles and presets.
____________________ Frog Leaping photobook: https://www.frogleaping.org
@Ulysseita: When I use the camera standard profile in Lightroom I get the remark "you can't use the camera standard profile! We specifically say on our website that DPP and the jpegs are emulated".
Ok so I use DPP to proof what the issues are, because it's a VERY similar test scene with very similar lighting.
But now even that isn't good enough. All excuses to blame it on the user.
So I show the BIG difference in the greens, all in similar light.
Yet again this isn't enough.
Now I see that same example you've used before, with the camera standard profile selected in LR.
+ it's an example of the 5DIV, a profile for which I confirmed (in the same post earlier today) doesn't have the big problems with the greens.
Mitch Alland wrote:
@Jochenb@ - Do you think that you could turn the emotion down, say 10 notches? I know English is not your native language, but you're coming across as screaming — and, as @Ulysseita@ states, you're not comparing like with like. I am writing this because I feel that this thread is important for many photographers because, as I see it, Cobalt-Image offers the most accurate profiles that can be purchased (as opposed to making one's own) — and the simulation presets are of great general interest. Of course, you should provide your comments and suggestions, which can be helpful for a software developer, but, in the context of this thread, my suggestion would be for you to take this particular discussion to PMs and, then, only post here what is comparable or the conclusions.
@Fred Miranda@ - I'm grateful to you for starting this valuable thread. You may want to consider deleting the last few exchanges that I've referred to for the sake of clarity, so that this thread can remain useful for photographers who are interested in the Cobalt-Image profiles and presets.
____________________ Frog Leaping photobook: https://www.frogleaping.org...Show more →
Oh I'm not emotional about this
It's a forum so words can often be misinterpreted.
Let's also be very clear again: I really appreciate Cobalt's work. The base profiles are great.
What I just don't like about this whole discussion is when I take the effort to explain (with multiple examples in similar conditions) what are still issues in some profiles it keeps getting discarded as "user error". I'm not that big of a fool.
Some emulations are well done, some just not yet. Why can't I mention this if their website literally says it's "a perfect emulation of the DPP output"?
Why also ask Fred to delete my posts?? They are completely ontopic. They might even be very helpful for other people that might be interested in that for example R5 emulation.
Jochenb wrote:
When I use the camera standard profile in Lightroom I get the remark "you can't use the camera standard profile! We specifically say on our website that DPP and the jpegs are emulated".
Ok so I use DPP to proof what the issues are, because it's a VERY similar test scene with very similar lighting.
But now even that isn't good enough. All excuses to blame it on the user.
So I show the BIG difference in the greens, all in similar light.
Yet again this isn't enough.
Now I see that same example you've used before, with the camera standard profile selected in LR.
+ it's an example of the 5DIV, a profile for which I confirmed (in the same post earlier today) doesn't have the big problems with the greens....Show more →
Let's see this difference on the green on the old post I made comparing R5 cobalt with DPP r5, and, I remember, different raw decoding means different curves and final results, so we should keep out all the different variables from colors.
If you don't think that and you assume: the color emulation whatever camera, lens, light condition, software, illuminant SCENE, has to be perfect, could be a problem.
So, reposting is the only thing I'll do if the conversation will follow in that mood.
"
We fixed the Canon Contemporary emulation pack.
Now the list of cameras has changed from:
Canon 1dxmkII
Canon 5d mkIV
Canon 5dsr
Canon Eos R
to
Canon 1dxmkII
Canon 5d mkIV
Canon 5dsr
Canon Eos R5
so in the next hours who bought this pack will find on his personal area on the website a new version to download.
How about the color accuracy?
The first step is to know that: DPP and LR have a different raw engine for saturation, exposure, etc.. these samples have NOTHING added after the opening apart from the Daylight WB.
these files are R5 raw.
On top the COBALT R5 emulation made from Lightroom
Below is the Canon DPP version of the R5 original file.
Saturation is external from the color profile and I only see this small difference; please do not forget the wider color reproduction level of the emulation built over the Cobalt profile.
Last but not least, with that pack you can have again on Lightroom the "camera matching" profiles made by us, not anymore available since the release of both R5 and R6.
"
that means.
1-camera matching is the emulation made by adobe over DPP to compensate for differences between software raw decoding and having the same colors.
2-Adobe is not doing anymore camera matching profiles for Canon R5-R6 so WE CAN'T use for Cobalt and our origin the same software and some little differences are impossible to manage.
3-"VERY similar" "similar" on color is ... a bad starting point to go straight into conclusions.
Jochenb wrote:
...Why also ask Fred to delete my posts?? They are completely ontopic.
I'm not asking Fred to delete just your posts, but to delete the whole interchange, including my last post. I feel strongly that this thread is valuable for people interested in these profiles: it has great examples and a lot of explanations that are valuable for users struggling with color correction and simulation. Inevitably, it's going to be long. The last interchanges between you and Ulysseita can be better handled in PMs — and I am sure can be resolved that way — without readers having to try to figure out whether you're making a valid comparison or not, or whether Ulysseita is hiding behind "user error" as you put it.
____________________ Frog Leaping photobook: https://www.frogleaping.org
@Ulysseita, yes those examples are all nice, but they don't show what I'm discussing.
You are comparing the DPP output to the output of LR with your profile on the same file, from the same Canon camera.
I'm applying the Canon emulations on Sony RAW files and that's where things look off. Just like I also showed with the M9 DNG emulation on a Sony file.
You then made the same comparison, again on the same file from the same camera where indeed: things looked much closer.
I thought the whole idea of the emulations was also to be able to "mix and match" cameras from multiple brands while maintaining the same looking output from all?
Or am I getting this part wrong and does that only apply for your base profiles?
Jochenb wrote:
You then made the same comparison, again on the same file from the same camera where indeed: things looked much closer.
I thought the whole idea of the emulations was also to be able to "mix and match" cameras from multiple brands while maintaining the same looking output from all?
Or am I getting this part wrong and does that only apply for your base profiles?
no, you're right.
That's what we did for the sample on A9 and 5dIV...and you never did.
I am still waiting for samples of the same scene with the same light of different cameras to analyze; whit that I would be more than happy to find how to improve our product.
I am not sure about the exchange, but I dó see a lárge difference in the greens with the samples Jochen posted. So a valid critique as far as I can see that may be of worth to other Canon R5 users.
Ulysseita wrote:
no, you're right.
That's what we did for the sample on A9 and 5dIV...and you never did.
I am still waiting for samples of the same scene with the same light of different cameras to analyze; whit that I would be more than happy to find how to improve our product.
You're not answering my key question Ulysseita.
That's what might be making this discussion confusing for some.
The examples you show are only interesting for users that want a profile that matches the output of the manufacturer.
For example an EOS R5 user that wants to have the DPP output in LR.
As you showed, you did a nice job on that.
However, we are talking about different things. I'm more interested in getting that Canon DPP look on my Sony files. That's the whole point.
That's why I asked this:
"I thought the whole idea of the emulations was also to be able to "mix and match" cameras from multiple brands while maintaining the same looking output from all?
Or am I getting this part wrong and does that only apply for your base profiles?"
I see these quotes on your website so I don't think I got that wrong:
"Additionally, if you feel creative and adventurous, have fun experimenting with different shooting experiences with our emulations, feeling like your camera can turn into whichever brand you always wanted to try out."
and
"Enjoy the Canon Contemporary style on all your cameras!"
@Mitch Alland: This is a discussion forum, with an ontopic discussion.
Because Ulysseita is very active in this thread I don't think it's inappropriate to discuss these things here.
Why do you feel these posts are ruining this thread if I may ask?
Is it only allowed to be positive about everyting? I'm very positive about some of Cobalt's work. I even said that the base profiles are the best you can get on the market, there's no higher praise I can give them.
Jochenb wrote:
"I thought the whole idea of the emulations was also to be able to "mix and match" cameras from multiple brands while maintaining the same looking output from all?
I see these quotes on your website so I don't think I got that wrong:
"Additionally, if you feel creative and adventurous, have fun experimenting with different shooting experiences with our emulations, feeling like your camera can turn into whichever brand you always wanted to try out."
and
"Enjoy the Canon Contemporary style on all your cameras!"
yes, and this is what happens when you take a picture with a Ricoh GR, and a Canon 5dmkiv of the same scene, and you use for the Ricoh the Cobalt 5div std and for the Canon the Camera Matching STD.
Different cameras, SAME canon colors when you compare comparable pictures from different cameras.
Ok if you want to proof me I'm wrong about how it behaves on Sony files, please show me comparisons with Sony. Also: show examples of the specific color range I'm discussing.
(BTW, the pink car under the chair also looks different in your example. The same goes for that small pillow that's more on the purple side on the Canon shot, while it's more blue on the Ricoh with Canon profile. And so on. So you show it yourself that it's not a "perfect" emulation.)
I have a sample that's exactly what's been asked for. The same scene shot with the R5 and A9, not even a minute apart so exactly the same conditions:
Canon DPP standard:
Sony with R5 standard profile in LR:
This result is again in line with all the other examples I've been showing.
Don't worry, I'm giving up after this. I've done all I can to make it clear.
Archive 2021 · Cobalt Profiles
