When I moved my main machine to a retina Macbook Pro, I was surprised to notice that the screen made my website images appear blurry. I had thought image sharpness would stay the same (or improve) since the physical resolution of the screen was much higher. I also thought that any resampling by the browser wouldn’t affect sharpness since there was an integer scaling factor: i.e. each image pixel should get mapped directly to a 2×2 square of retina pixels.

My assumptions were clearly wrong though as my web images were noticeably fuzzy.

To understand what was happening, I viewed my images in different browsers (Chrome/Safari/Firefox) and took retina resolution screenshots to see the pixel level detail. My test image was a picture of the Cliff Palace ruins in Mesa Verde and was produced with Lightroom (resized for web and exported with sharpening for screen).

This image of Cliff Palace may appear blurry on retina screens.

This image of Cliff Palace may appear blurry on retina screens.

Below is an expanded view of the upper window in the left tower based on screenshots with the retina display set to best quality (i.e. a straight pixel doubling).

Safari, Chrome, and Preview (Mac native viewer) of non-retina images on a retina display.

Comparison of Firefox, Safari, Chrome, and Preview (Mac native viewer) of non-retina images on a retina display.

In terms of sharpness Preview > Chrome (v39) > Safari (8.0.2) = Firefox (v34). I think Preview is applying a straight mapping of 1 image pixel to a 2×2 square of retina pixels. To see this, download the image and magnify it until you can see the individual pixels as blocks — the Preview screenshot will have pixels that are 4x the size of Chrome/Safari.

From the expanded view, I believe all three web browsers resample the image (probably with bicubic interpolation) and this blurs detail. Chrome tries to recover by applying sharpening but this introduces some artifacts as you can see a halo around the window. However the halo is not visible at normal magnification and I think it is a reasonable policy compared with Safari/Firefox which just leave the image slightly blurred.

What does this mean for my website? Retina and hi-dpi screens are currently only a small slice of my web traffic so I could feasibly do nothing and live with the blurriness. However, I’d like to get ahead of this issue as hi-res screens are only going to become more popular. This means I need to either upgrade to retina resolution images or find some way to force Safari/Firefox to apply additional image sharpening similar to Chrome (perhaps through a CSS effect if available).


The OWC ThunderBay 4 is a Thunderbolt 2 hard disk enclosure that holds up to 4 drives. The enclosure does not have hardware RAID support (although this is possible with software) and the disks appear individually to the operating system. In other words this is a JBOD (“just a bunch of disks”) enclosure.

ThunderBay 4 - a four bay Thunderbolt 2 hard disk enclosure.

ThunderBay 4 – a four bay Thunderbolt 2 hard disk enclosure made by Other World Computing (OWC).

Note that the “ThunderBay 4″ is an updated version of the “ThunderBay IV” enclosure. The primary (and perhaps only) difference is that the enclosure now supports Thunderbolt 2 connections.

I purchased my ThunderBay 4 because I was tired of juggling individual drives in my USB dock in conjunction with my Macbook Pro. Previously I used a Mac Pro with 4 internal drive bays but I had switched to a laptop for portability.

I ordered the 12TB version that includes four 3TB Toshiba consumer drives (model number DT01ACA300). The enclosure can also be purchased in other drive configurations as well as diskless.

Ordering from Other World Computing

I ordered the ThunderBay directly from Other World Computing (OWC). Ordering was straightforward on their website although I needed a user account to purchase (very annoying). My shipping address was different from my billing address and an OWC service rep gave me a call within an hour to confirm.

The free shipping was extremely fast. I had ordered on a friday afternoon and the package was delivered by the middle of the next day. It helps that I was only one state over from Illinois but I had still expected at least a day for OWC to process the order before the package was ready for Fedex.

The ThunderBay was double boxed and it was the best packing job I have ever seen (much better than B&H or Amazon). Everything fit snuggly within the delivery box with no possibility of slipping.

The 4 hard disks came already mounted in their drive sleds but they were not installed in the enclosure. Instead they were wrapped in an anti-static bag and surrounded by an extremely sturdy “air mattress” which kept them separate and away from the box walls. The plastic was very thick unlike the thin air bags that come in Amazon boxes that are easy to tear.

Packaging for the ThunderBay 4 enclosure.

Packaging for the ThunderBay 4 enclosure.

Installation and Setup

Very easy. Basically, open the front grill with the enclosed keys and slide the drives into each bay. Lock the drives into place by hand tightening a mounting screw at the top of each bay. Note to manufacturer: please ditch the keys needed to access the drive bays — it doesn’t add any security and it’s just another item that I need to keep from losing.

I used Apple’s Disk Utility to do a full format of each drive (to do a full format, erase the drive and use the security option to write zeros over every bit). This takes about 5 hours to do one 3TB drive but can be done in parallel by opening multiple copies of Disk Utility (in terminal type “open -n /Applications/Utilities/Disk\ Utility.app/”).

I also used Disk Utility to setup the device in various configurations (individual disks and RAID 0 sets) for testing.

Fan Noise

I’m sensitive to noise and online reviews of the ThunderBay suggested that the drive is very quiet (an important factor in my decision to purchase it). However, I found that the ThunderBay’s fan is very noticeable and I was disappointed that it was not quieter.

Without having a sound meter, I’ll describe it as being a little less loud than a Macbook Pro or iMac with the fans running full tilt. In a quiet room you will hear the fan unless you put the device far away. It’s nowhere near as quiet as a Mac Pro (which is silent to my ears). However it is much better than my old ReadyNAS which was unbearably loud.

As far as I can tell, the fan seems to run at a constant speed regardless of the load on the hard drives while the computer is on and connected. However, if the computer is sleeping the fan will eventually shut down. The hard drives power down when not in use.

BlackMagic Disk Speed Test

I tested the ThunderBay in various configurations using the BlackMagic speed test with 5GB files. For comparison, I also included results for the internal PCIe Flash storage on my Macbook Pro (PCIe flash is essentially a faster SSD drive):


Note that this test was done with disks empty (hard disks slow down as data is added). The results are pretty much what I expected: a single drive is comparable to speeds in an internal bay of my Mac Pro, and multiple drives in RAID 0 set scale not quite linearly. With 4 drives in RAID 0, it’s about as fast as the internal flash memory on my laptop.

My Setup

Before running further tests, I setup my ThunderBay in the configuration I planned to use it:

  • 3 disks in a RAID 0 array for my image files
  • 1 independent disk encrypted for storing other data

Lightroom and Photoshop Speed Tests

I copied my image library (~2TB) to the RAID and I then ran a number of tests using Lightroom and Photoshop. For comparison, I’m also going to report times for

  • a 3TB Hitachi drive running on a USB 3 dock
  • the internal PCIe flash storage on my MBP (this is also my boot/system disk)

For the following tests, all timing results for the Thunderbay 4 are obtained from the 3 disk RAID 0 set. Timing was done by hand and is probably accurate to within 1/2 second.

Test 1: Export RAW Files

I used Lightroom to process 111 RAW files from my Canon 5d Mark II (about 25-30MB each) and export 16-bit tiff files (170 MB each). As might be expected the internal flash memory was the fastest followed closely by the ThunderBay 4. The USB3 drive was 50% slower than the internal flash drive. However the gap is surprisingly small compared to the BlackMagic results.


Test 2: Open/Save Large Flat Tiff File

This test involved opening and saving in Photoshop a large uncompressed TIFF file (1.8GB) with just a single layer. In this case the results mirrored the BlackMagic times and both flash and ThunderBay were about 4x faster than the USB.


Test 3: Open/Save Large Layered Tiff File (RLE compression of layers)

This test involved opening and saving a large TIFF file (2.7 GB) which is a composite image made from 3 stitched RAWS with multiple layers added in photoshop for processing. The base image is uncompressed but the layers were compressed with RLE.

From the times we see that the ThunderBay is comparable to the internal flash memory. It actually comes up a little faster on the open but this may be due to measurement variance. USB3 takes 3 to 4 times as long.


Test 4: Open/Save Large Layered Tiff File (ZIP compression of layers)

This test involves the same file as test #3 but using ZIP layer compression instead of RLE. ZIP compression saves significant space as the file is only 1.5 GB instead of 2.7 GB, but increases processing time during open and save operations.

From the graph we see that there is almost no difference between the three storage devices. In this case, CPU processing is the limiting factor not the storage I/O bandwidth. Given that disk memory is cheap, I would avoid using ZIP compression if possible.


Temperature tests

For this test, I converted several thousand tiff files to jpegs in Lightroom and then measured the drive temperatures after approximately 30 minutes have passed (while the conversion was still ongoing).


Overall the drives stayed cool and are well under the manufacturer’s operating limit of 60C. As I measured the temperatures repeatedly, I noticed that the middle two drives were usually 1 degree cooler than the outer two disks.

Bundled Software

The enclosure is supposed to come with the following bundled software:

  • Prosoft Engineering Data Backup 3 (mac)
  • Intech SpeedTools (mac/win)
  • NovaStor NovaBackup (win)

My package didn’t have a CD but rather a slip of paper with a URL link to where the programs could be downloaded. However they were not at the stated location — I emailed customer service and they sent me a working link within a day.

SMART Status

It is possible to read the SMART status of drives in the ThunderBay. I used the open source SmartMonTools (http://www.smartmontools.org) to read information from each of the drives in the enclosure. (Note: although SmartMonTools is free, you may need to compile an executable from source code.)

The bundled software (Intech SpeedTools) can also read the SMART data but doesn’t appear to show the raw values of attributes.


Update 8/27/14

After using the ThunderBay 4 for two months now, I’ve noticed that it occasionally disconnects and may cause my computer to hang (necessitating a hard reboot). I haven’t tracked down the cause but it usually occurs when I’ve got the ThunderBay connected and powered on but not actively accessing the disk.

The disconnects and forced reboots are very annoying but otherwise the device works great. Lloyd Chambers at macperformanceguide.com believes this may be due to loose/poor fit on the thunderbolt port on the mac end. I’m not convinced yet that this is my problem but I plan to test this hypothesis (by getting an official Apple cable and working with several computers).

Update 11/07/14

Due to extended travel, I never purchased an additional thunderbolt cable to track down my disconnect problem. However, I’ve upgraded from Mavericks (10.9) to Yosemite (10.10) and have not had a disconnect yet. As of 11/26, I’ve still not had any disconnects so I think it’s safe to assume this has fixed the problem.

Some Concluding Thoughts

  • The ThunderBay 4 is expensive for a JBOD box that just holds drives. However alternative Thunderbolt enclosures (e.g. Caldigit T3 / T4, Lacie 5Big, DataTale) are just as costly. Thunderbolt enclosures with hardware RAID are even pricier.
  • In terms of speed, the drive enclosure is terrific and met my expectations for throughput. I have my main image library setup on a three drive RAID 0 array and this performed very similarly to the internal PCIe flash storage on my MBP.
  • In terms of noise, I wish the drive was quieter but the noise level is acceptable and many enclosures are worse.
  • The enclosure supports SMART status so that you can monitor the health of your drives. Given that very few Macs have internal bays, this is a very useful feature (SMART is not supported on USB for macs).
  • Hard drives are kept cool in the enclosure and temperatures did not exceed 35 degrees Celsius in my tests.
  • I’ve experienced disconnect problems with the ThunderBay 4. I’m still trying to track down the cause and if there is a simple setting fix.


Serendipity at Santa Monica Beach

April 30, 2014

I normally think about photography in terms of making pictures, where one carefully plans out the composition and light, but sometimes you just get lucky and a photo just appears in front of your camera. For this picture, I had setup on Santa Monica beach with the intention of photographing the pier and ferris wheel […]

Read the full article →

Last Night in Prague

April 19, 2014

I took this photo on my last night in Prague. Before visiting the city, I came up with a shooting list and I knew that I want to create some iconic images of Prague’s bridges. I had already shot the Charles bridge but hadn’t yet seen a good location where I could have multiple bridges […]

Read the full article →

File Naming Systems for Managing Your Image Collection

April 16, 2014

Once you have more than a handful of images it’s going to be very important to use a systematic naming scheme. Having an organized system for numbering your files will allow you and your clients to identify images precisely. Instead of a client asking for a picture of that “prison cell in Alcatraz” (I have […]

Read the full article →

Baby Sea Turtle Struggles to Reach the Ocean

April 9, 2014

Went I went to Costa Rica, one of the events I was very lucky to witness was the hatching of baby sea turtles on Tortuguero Beach and watching them make the long crawl from their nest to the ocean. Tortuguero is the largest Atlantic nesting site of the endangered green turtle. Green sea turtles nest […]

Read the full article →

An Edible and Very Delicious Camera Cake

April 5, 2014

The cake my wife got me for my birthday: It’s all edible including the camera body, lens, and even the book covers (I think these were made with some type of edible ink).

Read the full article →

Silverfast Resolution Target for Measuring Scanner Performance

March 30, 2014

I recently borrowed a Silverfast resolution target to measure the performance of my scanners. The target is a slide that is printed with a series of black bars which get progressively smaller and closer together (the pattern is based on the USAF-1951 test pattern). The slide costs approximately $60 at B&H Photo. You scan the […]

Read the full article →

A Lightroom plugin for verifying images with hashes

April 21, 2013

For the past two months, I’ve been working on a plug-in to verify image files in Lightroom. The plugin works by computing a hash for each file which can be thought of as a digital fingerprint. As long as the image is unchanged (i.e., the file is still ok) the hash value will also remain […]

Read the full article →

Numbers behind a stock photo and print business

January 6, 2013

In the spirit of Harper’s Index here are some random facts about my photo business: Largest number of prints in a single order: 11 Largest number of images licensed in a single transaction: 12 Biggest reproduction made from my work: 8’x25′ (400 sq. feet) Biggest license: $5,840 Biggest license for a single image: $1,400 Smallest […]

Read the full article →