Back in October 2016, I reported on broken Chrome color display in Advisory: Google Chrome Color Management is Broken, at least on macOS.
Jonathan N writes:
After recently upgrading to a iMac 5K, with a wide gamut monitor, I discovered that the Chrome browser’s color accuracy is broken, as described in your blog post.
As you describe, untagged images and CSS colours are incorrectly rendered in the monitor’s color profile, instead of sRGB.
After much digging through Chromium bug reports I was happy to discover that this has been fixed in Chrome Canary (the bleeding edge version of Chrome). You may already be aware of this, but I am writing to let you know, in case you want to test for yourself.
The relevant fix is described here. To get it working you need to set Chrome’s --enable-color-correct-rendering flag to “enabled”. In Canary visit chrome://flags to access the flag. This flag is only included in Canary, not the stable branch of Chrome.
After enabling the flag untagged and tagged images on your color rendering test page now match! CSS colours, which were previously horribly oversaturated now also appear to render correctly.
Many thanks for the information you provided about this bug on your blog.
DIGLLOYD: I got to great lengths to deliver just the color I want when I present my images, so I am glad to hear this bug will likely be fixed in the regular release at some point. I do not like my work to be seen incorrectly. Heck, I don’t even like it to be seen on inferior displays—it just looks so much better when viewed properly.
I urge all of my macOS readers to use Apple Safari. Safari is not bug free, but it is the fastest browser for displaying large images (particularly my aperture series toggles), and it has the fewest color display bugs. Particularly if using an iMac 5K (the best display for viewing images at any price as of mid 2017).
This series looks at sharpness across the field, along with foreground and background color correction in out of focus areas.
In my review of the Sigma 135mm f/1.8 DG HSM Art in Advanced DSLR:
Sigma 135mm f/1.8 DG HSM Art Aperture Series: Pescadero Creek Downstream (Nikon D810)
Includes images up to full resolution from f/1.8 through f/13.
David C writes:
That shot of pescadero creek is amazing. please don’t tell me it was handheld
DIGLLOYD: it’s not possible to handhold an aperture series, and I would never evaluate sharpness handheld at 1/13 second.
Eric B writes:
Michael Tapes LensAlign Tool
You are often testing new autofocus lenses, and often on the D810 DSLR. Do you focus tune these lenses before you shoot them and if not, why not.
I am now a mirrorless user but tried to focus tune my autofocus lenses when I used my D800E. I found it challenging to say the least. I’d appreciate hearing your comments about focus tuning and how you choose to focus manual focus lenses when you test those.
DIGLLOYD: See the lens align articles I’ve written over the years. Michael Tapes LensAlign Tool comes in various flavors from about $125 to about $85. If you’re going to focus tune your camera, it’s the best tool I know of.
I do not focus calibrate my lenses for two reasons: (1) I have had little trouble, and (2) I cannot present work on sharpness assuming that autofocus worked optimally: even very small errors would invalidate conclusions and gyrate areas away from the point of focus into total confusion as to peak performance (think field curvature and focus shift, which vary across the frame).
Therefore I do all my focusing in magnified Live View anytime I am evaluating image sharpness or making a comparison. Anyone relying on autofocus when assessing lens sharpness is incompetent, unless the goal is to assess sharpness of the system as a whole (what one gets on average using AF with any given lens, which often leaves out the lighting and subject matter thus making it even more lacking in rigor). Given variations in cameras and lenses and thus AF errors, nothing else has any rigor at all. I assess the peak optical quality, and that is hard enough to get right even in magnified Live View.
Alignment, precision and accuracy
To understand alignment, one must understand the what precision and accuracy mean, along with the damage from to sharpness from focus shift and that’s for starters.
If a camera does not compensate for focus shift and a lens has significant focus shift, then autofocus is hopeless for optimal results, since compensation is for only one aperture. So I’m going to ignore that here—just know it can be a huge factor for lenses with focus shift.
DSLR autofocus uses a separate optical path for the AF system versus the optical path to the sensor. This is a key reason why alignment may be needed—those two paths are highly unlikely to be much better than 40 microns in agreement, which is the difference between tack-sharp and visibly unsharp. Worse, the AF sensors as seen by the user are generally coarse in size and not always where they are shown to be! Let’s set that whole can of worms aside too.
The key issues with fine focus calibration are accuracy and precision:
- Suppose the target is at 3 meters... if the camera always focuses at 2 meters half of the time and 4 meters half of the time, it is thus deadly accurate—and every picture will be blurred and useless. Accuracy is an average deviation from the true value.
- Precision is critical: think of it as how tight the grouping. Ten shots fired into the bullseye are highly precise and highly accurate (our ideal scenario). 10 shots hitting the target but randomly spaced over the target is not very precise but is highly accurate—and that is pretty much what DSLRs do a lot of the time (in near/far deviation from actual target distance).
If precision is high, then the grouping is tight and fine focus adjust feature can be used to re-center that grouping for high accuracy in a tight grouping. But my years of experience with DSLRs prove over and over that DSLR autofocus tends to not be precise at all, and precision is terrible for lenses with less than outstanding contrast (like most of Nikon’s f/1.4 lenses), particularly at medium-far distance. Hence fine focus adjustment is worthwhile, but it cannot address a fundamental failure—lack of precision. All it can do is center the loose (imprecise) grouping somewhat better. Lenses with very high micro contrast generally do much better, f/2.8 lenses do better than f/1.4 lenses, etc.
As for mirrorless, Sony has a serious bug in focusing stopped down. One can see the magnitude of the error by initiating AF at f/8, then opening up to full aperture—there can be a great deal of error, plenty enough to blur the image badly at full aperture. It is a fundamental problem with Sony, which is why when I use Sony AF, I never focus at f/8 or f/11 (one can usually get away with f/5.6 for an f/2.8 lens but it’s no guarantee of optimal). Trying to focus stack at f/8 or f/11 with Sony is a time-wasting annoyance since the lens has to be opened up for every frame in order to focus properly. It’s a particular nuisance for wide angles.
All these nuances have taken me years to nail down. See Making Sharp Images and/or hire me for a day on a photo tour and I can go through it all in the field with you.
This series looks at the entire aperture range from f/1.8 through f/16. Of interest are peak micro contrast at f/1.8 as well as how well f/11 and f/16 hold up with diffraction mitigating sharpening.
In my review of the Sigma 135mm f/1.8 DG HSM Art in Advanced DSLR:
Sigma 135mm f/1.8 DG HSM Art Aperture Series: Pescadero Creek (Nikon D810)
Includes images up to full resolution from f/1.8 through f/16. Diffraction effects are evaluated and shown with/without diffraction mitigating sharpening.
This series and the American Hotel wail in protest for a 72 megapixel Nikon D900. I speak not of just resolution per se (surely a shortcoming of the D810), but also of sidestepping color aliasing and spurious resolution and color purity and similar digital limitations of a Bayer sensor that result when using extremely high quality lenses (presuming that resolution is only resolution is an fundamental conceptual error—see oversampling).
The about $1399 price for the Sigma 135mm f/1.8 is a screaming deal that awaits a higher-resolution Nikon camera body. Were Leica to ship this lens, Leica fanboys would rave about it and it would cost $8000. The performance here is better than any Leica M APO lens in resolving power and vastly better in terms of color correction. How Sigma can give make a lens this good at this price is surely intimidating to other companies.
Together with the Nikon 70-200mm f/2.8E FL ED VR (the zoom that looks like medium format in its rendering), we seem to be in a golden age for DSLR lenses, so much so that I question my Sony mirrorless inclinations, particularly if Nikon delivers a D820 or whatever it will be with higher resolution and at least as good dynamic range and per-pixel quality—and an EVF option in the hot shoe would seal the deal.