Sunday, June 22, 2014

Guest Post: Joseph Coco: Spyder3 vs. ColorHug, Colorimeter Showdown

If you enjoy the reviews on this blog, I strongly encourage you to consider donating!  All materials tested were purchased from my own pocket, using funds earned selling at conventions and filling commissions.  For guest posts and reviews, I usually try to find a way to reward my guest writers in some capacity, which still costs money.  If you don't want to donate, but would still like to help this blog continue to publish quality content, please consider making a purchase from the Natto-shop.  As of this time, Keep on Truckin' Nattosoup is not under any sponsorship, and all views are either mine or my contributors.

For digital artists, color accuracy on their monitor and printer settings can be vital in the creation of digital art.  Although different monitors display colors uniquely, a color-corrected monitor helps the original artist make informed decisions in regard to color.  Even traditional artists may find a color-corrected monitor to be an important tool in their arsenal, especially artists who digitize their pieces.  For me, a color accurate monitor is essential in the creation of my comic, 7" Kara.  7" Kara is a watercolor comic, and for it to be reproduced correctly, I need to make sure that every step of the digitization process is the best it can be.  This means investing in a decent monitor, a computer that can handle large file sizes, having a good large format scanner, and making sure my monitor is color accurate, so I can correct color inaccuracies in scans.  Joseph Coco kindly wrote this post after adjusting my monitor for me, since we both thought other artists might find the information within to be useful.


Gamut Coverage and Backlighting

Color is important to artists, but monitors are notoriously bad at reproducing accurate color and staying in a reasonable price range. Their color gamut is simply too small to accurately show an artist what they are actually creating. That being said, not all monitors are created equal, and you should consider looking at parameters other than size, max resolution, and refresh rate next time you're in the market. As far as color goes, most monitors will cover all the sRGB colorspace, but higher end models, such as IPS panels, will often cover the sRGB colorspace as well as a significant amount of the Adobe RGB colorspace. This is not a hard and fast rule, and I don't pretend to understand all the features which go into covering a colorspace. But generally speaking, LED backlit IPS panels cover more of the spectrum than TN or VA panel, and I believe outperform most CRTs. TN panels (the most common LCD display) do have advantages over IPS, so you should weigh how important color accuracy is over refresh rate.

Source
The majority of LCD displays are poorly lit by two, or maybe even just one  light source. This not only creates inaccurate intensities of colors, but across the plane of the monitor colors are inaccurate to different degrees. So viewing an image on one part of the monitor could appear different than on another part. So ideally, a monitor would be evenly lit by many consistently bright, consistently-spaced sources of light. That being, full array LEDs. As far as I understand, white LED back-lighting is the most common, but RGB LED is considered superior, but more expensive. But LED isn't black and white. There are some monitors which are edge backlit by LEDs, which suffer from the same issue as having only a couple sources of uneven lighting--especially for large monitors. While other monitors have locally-dimmable full arrays of LEDs, which I imagine can create much deeper darks and brighter whites.

Source. Single backlight, in all its glory. Unfortunately this is common in the realm of laptops.

Land of Defaults

Just because a monitor can display a lot of colors doesn't mean it is doing it accurately. In other words, if it's configured wrong for your lighting environment, you may have the full range of colors possible, but no color being displayed is what someone looking at the same image in real life would see. So you would think color calibration would be absolutely vital to artists. So that they can ensure people see their art the way they created it. But in the world of technology, default is king. And the average consumer isn't going to have a wide gamut. And they're not going to color calibrate their displays every two weeks to accommodate for subtle light changes. So it may sound counter intuitive, but it often helps to dumbify the perfectly colored image you just created when you're putting it on the internet for all to consume. You can bet corporations do this before dumping images onto their sites. Because the satisfaction of a perfectly crafted image isn't going to make sales. No, that image needs to look good on grandpa's display.

The way most artists accomplish this is by using two displays. That being, one monitor with all the bells and whistles which is color calibrated regularly, and another out of the box, as cheap as possible monitor plugged into the same computer, with default settings and no software installed.

The alternative to creating a color-inaccurate 'web-ready' image is to embed the ICC color profile you've created for your calibrated monitor into the image, this way a modern web browser or image viewer can accommodate for the differences between the two monitor profiles.

Photoshop saving for web viewing and embedding a color profile
The reason it's important to keep that properly colored image as a base is not only for archival purposes. Printers also often wallow in the land of defaults, but they have some nifty calibrating features. A medium- to high-end inkjet printer is chock full of sensors to accommodate for things like humidity, air pressure, and temperature to achieve a consistent electrostatic spattering of ink onto your paper. So while printer calibrators exist, and I recommend you invest in one if you're regularly printing at home and already have a monitor calibrator, they aren't as necessary. So the image you see on your monitor, if it is indeed color corrected, will be more or less what is printed. I've seen countless people blame their printers for printing far too dark when their monitor's brightness is cranked up to 11. In other words, you won't be able to achieve consistent printing results on a mis-calibrated monitor, and it's very difficult to diagnose a printer without an accurate monitor.

I can't speak to laser or dye-sublimation printers, but I imagine the same principles apply.

Becca's Monitor


After some research, I advised Becca to purchase Achieva Shimian Lite (QH270-Lite,) which covers roughly 75% of the Adobe RGB colorspace. Which is quite a feat given its price, response times, size, and resolution. You've probably seen the panel before as it is also in Apple's Thunderbolt monitor.
QH270-Lite doesn't quite fill the AdobeRGB spectrum, it's obviously missing small portions around the edges and a significant amount of the blue-violet spectrum.



I took a couple snap shots of the monitor's colorspace as recorded by the two colorimeter devices. These 3D LUTs were produced by Gamutvision.
If the two calibration devices were equal--given that the lighting was the same during both calibration cycles, then the colorspace should be equivalent. But the devices obviously have some variability.

Lighting

I've mentioned a monitor needs to be calibrated for lighting, and that you should change your lighting as little as possible. But while sunlight, with a color rendering index of 100, has the highest ability to accurately represent colors, it is extremely inconsistent. The properties of sunlight are constantly changing throughout the day, and to a greater degree, throughout the seasons. So while you should have some sunlight in your studio, you should also aim to have artificial light sources to get up to roughly 8000 lumens in your studio space.

Tungsten bulbs have a high CRI, but a poor temperature for color accuracy. There are many types of bulbs, but I can only speak to Tungsten, compact fluorescent, and LED. Generally speaking, if a bulb has a neutral temperature, full-spectrum CFL have a higher CRI than standard CFL, which have a higher CRI than LED bulbs. So next time you're shopping for bulbs for your studio, try to check the temperature and CRI stats.

Apparently 3 of the bulbs Becca is using in her studio have a CRI of 74, which is a bit on the low end, while one has a CRI of 82. With a total of 5300 lumens from lamps, and the rest coming from the one large window.

Spyder3

I bought Becca a Spyder colorimeter by Datacolor as a Christmas gift some time ago. It's graded to calibrate CRTs, LCDs, and laptops. I read about and wanted to use a color calibrator since I was experimenting with digital art in high school and was trying to convince Becca it was a good investment. I also wanted to show Becca I supported her career, and at the time she frequently worked digitally. Regardless, she does some work digital, regularly references digital images, and always scans her final product.

I chose to get a Spyder second hand because many people believe you have to calibrate your monitor once and then you're good to go. While this is mostly true, if any natural light is entering your room, or your light sources move or degrade with time, it's wise to recalibrate.

I went to install the software for Spyder since it has been a while and it requested a key. I'm positive I have paper sleeve it came in buried under paperwork, but it's not worth searching for it. I attempted to get the key emailed, but apparently I never registered the product. So I decided it would make the match between ColorHug and Spyder3 even if I used the same open-source software to calibrate it. That being DispcalGUI.


The Spyder came with a suction cup, but everyone advises against using these on LCDs as they can be destructive. I opted to rubber band it to the monitor in stead.


DispcalGUI 2.1 recognized the Spyder and was ready to go as soon as I turned it on. I didn't try to flash the firmware since that would likely require me to dig up the key they insist I have. Nevertheless, the results looked good. It took ~2 hours which seems unusually long, though I did not watch it. It ran through 3 iterations. On the third it took 309 measurements and had an average neutral error of 0.307448 deltaE. As I understand, the human eye can only differentiate 1-2.5 deltaE. So it seems the calibration was a success. You can view the entire log if you're interested in the details.

This shows how much the values which are to be displayed on the monitor were affected by the Spyder calibration.

ColorHug

I can't recall how I came upon ColorHug, but I'm a sucker for open hardware and it was cheaper than a Spyder. Now, why, you ask was I in the market for a colorimeter considering the Spyder was performing well?   I opted to purchase the ColorHug to support the project itself and for review purposes. I thought an affordable colorimeter would be a fantastic addition to any artist's arsenal, and Becca was extremely excited at the idea of reviewing competing models.


The ColorHug has a smaller form factor than the Spyder. It came with a strap rather than a suction cup, but I already had the rubber band, so why bother. I did flash the firmware... after I ran it once. The process was nearly transparent. I clicked a button and waited a minute, then went on with the calibration. But rerunning the calibration with ColorHug was probably for the best anyway because I took pictures of the ColorHug while it was actively being used which I believed botched the results. It is worth noting that the screen caps may have had some small effect on the calibration as well.
Initial adjustment of brightness and temperature to allow for easier color matching. You'll find you monitor is likely too bright, but temperature is all over the chart.

You'll see this a lot.


You should probably just walk away from your computer for an hour. But don't turn off the lights!
Ah. Complete, but with little fan fair. Out comes the ICC and a barrage of other files. If you want to better understand the ICC output you can read the 130 page ICC specification.
The first thing I noticed was that ColorHug had more options for the type of display I was calibrating. This may be because the developer worked with DispcalGUI directly rather than creating a propriety tool. The two main options were LCD (generic) and LCD (white LED backlit). Now here's where things got complicated. I knew the QH270-Lite was LED backlit, but I had no idea if it was white LEDs or RGB. I couldn't find the answer to this on the internet, so I assumed they were white LED because of the price-range of the monitor and went a head and calibrated. Unfortunately, the type of LED is not advertised if it's white because it's considered inferior. But upon looking at the above picture, it looks like the X in the photo was caused by the reflection of the flash of my camera and illuminated the LED backlights. They look like RGB... but to be honest I have no idea and I doubt it effected the calibration much.

The second calibration with the 1.2.1 in stead of 1.18 firmware of the ColorHug was a success. It took roughly an hour with 3 iterations, the third taking 328 measurements. The average neutral error was 0.320086, which is far below what the human eye can discern, but slightly higher than that of the Spyder. You can view the entire log if you're interested in the details.

This shows how much the values which are to be displayed on the monitor were affected by the ColorHug calibration.

Showdown

Spyder3 left. ColorHug right.
So which device fared better? Or more importantly, does it make a difference and should you just go for the cheapest colorimeter you can find?


I chose to include the botched ColorHug run just for completion's sake, but it wasn't an actual contender. All the images' colors were simulated using IrfanView's color correction configuration. In other words, I didn't change the monitor's profile and photograph the monitor. I don't have the original in front of me, but I remember the colors being vivid. To me, the ColorHug updated looks the most Copic-like, which is what was used to produce this scanned piece. But honestly, there isn't a significant difference between Spyder3 and ColorHug, especially considering the price difference between the two.

Looking at the dump of the stats from the ICC profiles, it seems like the only significant differences between them are the phosphor/colorant chromaticity, the metamerism index, and whatever DevD, CIED, and mmrod stand for.

Obviously different people will receive different results. It's possible your monitor is perfectly configured for your lighting, but unlikely. Or if you work on a laptop in different spaces all the time, it may not be worth the investment. I would recommend you borrow a colorimeter from a friend if you are concerned about the reproducibility or printability of your images. If you get positive results from it, and have around $100 to drop on a new art toy, definitely check it out. Or you can go in with some friends on one. I love the ColorHug, but Spyder and X-Rite are also options.

Price Breakdown:


ColorHug (http://www.hughski.com/buy.html): Approximately 85 USD. It can be purchased via Paypal or Bitcoin, but the site itself is not set up to directly accept credit cards.
Spyder3 (http://www.amazon.com/ColorVision-S2EL100-Spyder-3-Elite/dp/B000X4X35C): $111.11

I personally think the ColorHug did the better calibration job, although Becca disagrees and says it's the Spider. She does concede that the ColorHug did an excellent job, and the difference is almost unnoticable- she could only tell the difference because she's familiar with the original piece as a marker render, not just a digitized scan. Both are a huge improvement over the default.