R-D1 update

NOTE: The sample images have been moved onto sub-pages to speed the loading of this article.

What's changed

Well, I've finally managed to find time to update my Epson R-D1's firmware. With this update an R-D1 becomes a R-D1s in all but name - or so Epson claim. I don't have any reason to disbelieve them, but I also don't have an R-D1s to compare against. So I'm going to assume the claim is true.

What interests me are the effects on the operation of the camera and I have to say it seems to be almost entirely good news. Post-update the changes I've noticed are...

16x zoom on RAW images - previously you could only zoom in about 2 or 3 times, I believe you could do 16x on JPEG files but I have never used them as I take all shots in RAW mode. The improvement is very welcome as it is now possible to actually check the critical focus of shots in-the-field. When you flip into zoom mode the camera does some processing (which takes about 4 seconds) before displaying a zoomable version of the image. The processing seems to be speed driven as the actual zoomed image doesn't appear to be as good as the picture you get in Adobe Photoshop. However, it is perfectly acceptable for checking focus, depth-of-field, etc.
Improved RAW buffer depth - now this is wonderful. You can now take 3 rapid shots before you hit a stall due to buffering. Previously only 2 could be taken and a 50% improvement is amazing considering that there is no hardware update. Unfortunately, as you'll see below, the writing speed to the card has not improved in the same way.
Removal of the 'double press' issue - the R-D1 is supposed to be a 'shooting priority' camera. That is, when you press the shutter it should take a shot, regardless. This is complicated on the R-D1 because you have to wind-on but, even if you did, it would not always fire the shutter on first press. I'm almost certain that this only occurred if you had the LCD display on. The first press of the shutter would turn the LCD off and a 2nd was needed to take a shot. Obviously if the LCD was not on, the press would fire take the shot. The net effect of this was that you sometimes missed shots because of the delay in realising you needed the 2nd press. Thankfully this 'feature' has now been removed. Regardless of if the LCD is on or off, the camera will fire on the first press of the shutter (assuming it's wound on & turned on).
Quick preview - it's now possible to set the camera to display the just-taken image after every shot. It seems to do this before writing the image to the card (I'm talking RAW mode here) and so is pretty quick. The only drawback is that it's just the image. You can't have a histogram displayed and you will need to press the playback button to see that - which only works after the picture has been stored to the card (exactly as before).
Histogram changes - the histogram has been improved. It's now place at the bottom-right instead of slap bang in the middle of the image. It also has lines showing stops to allow you to better gauge how much under / over it is and set the exposure compensation appropriately. Unfortunately, it's still not possible to show flashing hi-lights AND the histogram at the same time. It would also have been nice to have the option of RGB histograms. Still I'm glad of any improvement and I will now set the camera to display the images in this mode. Which brings me onto...
Playback setting remembered - before the update the camera used to forget which playback mode you were in. If you set the camera to display the histogram on playback, you would have to reset it when you turned the camera off and back on. After the update the camera will remember this setting.
Improve direct-print quality - if you decide to print directly from the camera to a printer, the camera will now interpolate the image to a larger size for improved print quality. Not something I plan on using, but it's nice to know it's there just in case.
Full format now longer - I said it was 'almost entirely good news', well here is some bad news. For some reason a full format of all the cards I own now takes longer. Before the time was negligible, perhaps a second or two. Now it varies between 4 to 20 seconds, but on one card it's over 3 MINUTES! In the latter case a quick format was almost instant, but I'm unsure about the storage speed implications of only doing a quick format. It may mean that repeated quick formatting slows streaming of data to the card. This latter issue doesn't really effect me and so I've not tested it. The time variance in the formatting times seems to be card type and card usage dependent. On my SanDisk Extreme III, 20 images took 4 seconds to format away and an almost full card (950Mb+) took 20 seconds.
Long exposure noise reduction - This is a welcome addition, removing stuck pixels in shots, but the problem with this mode is that it seriously effects the operation of the camera. After each shot is taken the camera locks up for slightly longer than the period of the exposure. You can't do anything with it - nothing. Fortunately, you can switch this noise reduction on and off.
Dead pixel correction - you can now force the camera to remap the dead pixels on the sensor so it can blend the stuck pixels out of images. I'm not sure you'd want to do this on a regular basis (I certainly hope I won't need to) but it's a useful feature to have.
RAW settings - you can now set the RAW mode to store just a RAW image or a RAW image and a JPEG. Obviously the latter is going to take more space on the card.
Colour spaces - you can now set the colour space for JPEG files (it doesn't effect RAW) to either sRGB or the larger Adobe setting. By using the latter you could get a broader range of colours in your JPEG file but you will need to use a colour space aware product, such as Photoshop, otherwise the colours will look out in your prints and on-screen. Again, I don't shoot JPEGs with this camera so I have no use for this feature.

Upgrade notes

There are a couple of things to note when doing an upgrade. Firstly, you need to read the instructions before doing it - especially the bit about making sure your battery is fully charged. Secondly, and this isn't clear in the instructions, you need to reset the camera settings and do a dead-pixel correction after the update. Initially I failed to do this and noticed some quality issues in the shots I took before doing so.

You can download the update from here - the instructions (in several languages) are included in the download. You may need the Adobe PDF reader (the Apple Mac with OSX 10.4 will just read them, the PC might). You can also download the new user manual from this page.

Image quality

This was the main concern of mine for the one-way-only update. There's no way to go back to the original firmware (I suppose you could ask Epson and pay) so if the updated camera took worse pictures I'd be very miffed. This was further compounded by the fact that no-one (that I could find on the web) seemed to have look at this issue.

Shadow noise has always been a bit of an annoyance for me, because if you under exposed a shot and have to push the exposure to get a decent print, it would introduce huge amounts of extra noise. My Canon EOS 1Ds (mk1) is massively better in this respect and you can push exposures by 1 or two 2 stops without causing major problems.

Even if you get the exposure right the vignette on some lenses, ie the Voigtlander 12mm f5.6 @ f5.6, would cause problems on the R-D1. You can correct for pretty much all of the vignette on this lens using Photoshop's Camera RAW but doing so often introduced noise into the edge of the shot. It becomes a balancing act between evenly lit shots or noisy corners. This occurs because the vignette removal in Camera Raw selectively brightens the shot - ie in the corners and therefore makes the noise in those areas more obvious.

So the question I wanted to answer was if a shot was slightly under exposed and pushed in Camera Raw, would there be less noise in the shots after the update?

Long exposures

Prior to the update I took a series of shots, including some of a bookshelf in our house. These are all about 1 stop under exposed and all long exposures. One shot was taken for each ISO rating on the camera. The camera was mounted on a tripod and I held the shutter open using the 'B' setting, timing the exposure on my stop watch.

Post update I re-ran the test as closely as possible, both with the Long Exposure Noise Reduction (for here on referred to as LENR) on and off. You can see the sample images for long exposure by clicking here (lots of large images).

Looking at these shots I have to say that there really is no difference in noise levels after the update. There is some variance in the exposures and this is due to the time difference between the sets. I had to use natural light (as I had no studio lighting available) and it changed between exposure sets. I would point out that the exposures are within a quarter stop of each other. The time exposure time and f-stop are the same as I adjusted the amount of shading in the room to get as close as possible on the exposures.

There is kind of good news though, because the Long Exposure Noise Reduction (LENR) had nicely removed the speckles and blotches seen in the pre-update and post-update, non-LENR shots. I assume this mode uses dark-mask subtraction to detect these blotches / speckles and then remove them. The only disadvantage is that after taking the shot, the camera takes slightly longer than the original exposure to do this operation. During this time the camera is completely locked-up and you can't do anything with it. In the case of a 34 second exposure, you'll be waiting about 40 seconds for it to complete.

IR shots

The R-D1 is a fairly capable camera for IR (inferred) photography because the filter in front of the sensor (which removes IR) isn't too strong. The Canon EOS 1Ds (mk1) has a very strong IR filter in front of the sensor which renders it pretty useless for this kind of photography.

The R-D1 filter is so subtle that you can actually take shots hand-held, this combined with the fact that the rangefinder allows composition with the B&W 92 IR filter fitted, makes the camera pretty ideal for IR work.

The update hasn't made much difference to the noise displayed in IR shots. There is a difference between the pre and post shots because the weather conditions changed and it became sunnier when the post update shots were taken. Having Long Exposure Noise Reduction (LENR) set to 'on' doesn't make any difference either - but then they weren't really long exposures so that's to be expected.

You can see the before and after IR shots here (lots of images) and you can see a better set of IR shots here (lots of images).

Standard shots

Given that the last two test for noise have not shown a reduction after the update, you won't be surprised to hear that the same is true in standard shots. Post-update, there is no real difference in the noise profile of the sensor. To see the test images, click here (lots of images).

Noise removal

It should be noted that the noise you see in the images has been made deliberately worse. Anyone looking at this to try and gauge what the noise is like on the R-D1 is likely to now think it's awful. This simply isn't the case and any noise that is there can be dealt with quite effectively in Camera Raw. For example images, click here (lots of images).

Card performance

The original article on card performance can be found here, the new test was conducted the same way (except I only did 2 runs). The overall storage times for 20 shots are faster after the update in some cases, slightly. This could be due to the extra buffer slot (but the camera still has to store the same amount of data) or the streaming speed has been ever-so-slightly improved. Equally it could be down to variability in the method of testing.

Card	Run 1	Run 2	Run 3	Average	Data rate	Frames / sec
Puremedia	93.4 secs	93.2 secs	93.2 secs	93.27 secs	2.04 Mb/s	0.21
After update	91.7 secs	91.2 secs	--	91.45 secs	2.08 Mb/s	0.22
Lexar	81.2 secs	81.2 secs	80.9 secs	81.10 secs	2.34 Mb/s	0.25
After update	81.0 secs	81.0 secs	--	81.00 secs	2.35 Mb/s	0.25
Extreme III	77.1 secs	77.0 secs	77.3 secs	77.13 secs	2.46 Mb/s	0.26
After update	76.7 secs	76.7 secs	--	76.70 secs	2.48 Mb/s	0.26

With the 'quick preview' set to 3 seconds (in the above tests it was off) there appears to be no difference in the time taken for the 20 shots, but I only tested this on the Extreme III card.

There is, however, the issue of full-format times taking longer. It used to take a second or two, I never actually timed it because it wasn't a problem. It now noticeably longer, seemingly card and data size dependent. I didn't do a lot of testing but here are the results I got. The time for the Lexar card is two hundred and five. Each test was run more than once and proved consistent on each card / data size.

Card	190Mb (20 shots)	950Mb
Puremedia	4 secs	--
Lexar	205 secs	--
Extreme III	5 secs	20 secs

The upshot from this is that I would either use a recommended branded card (such as the Extreme III) or, since the performance of this card is not fully used in the R-D1 (and the price is high), test the card's full-format time before you buy. I say a branded card, because some of the rebrands change supplier or use multiple suppliers, so the card's performance may not be consistent from one example to the next. Extreme III cards (in my limited testing) do seem to be consistent.

You could just use a 'quick format' which is almost instant on all the cards tested. However, my previous test showed that the streaming speed could be seriously effected if you don't do a full format between each test (on some cards). So using the quick format may slow down the shooting speed of the R-D1. You would need to test this, I've not.

Other reported issues

There is only one real outstanding concern that I have. There have been reports that, post-update, the R-D1 uses more power and the batteries don't last as long.

Personally, I purchased 3 batteries because of reports that the camera (in it's original, pre-update form) used batteries very quickly. Since then I've managed to get probably 200-300 shots out of each battery and have never really needed more than 2.

I don't regard this as heavy usage and I'm taking comments about the update increasing usage with a pinch of salt for the time being. I'll report back, if I find a problem, when I've used the post-update camera more.

Conclusion

All-in-all this is a fantastic update and it makes a very good camera even better. If you own an R-D1 then I would heartily recommend that you upgrade and receive all of the benefits of an R-D1s. Other than the formatting issue (and a possible battery issue) there really is no reason not too.

It's a shame that the image quality was not improved more, but I suppose the main issue here is the very old sensor technology in use and even the best firmware update can't change the hardware.

If you would like to buy an R-D1s then take a look at Robert White's website. They have started to supply this new model.

Update

I have found that the statements about battery life (this time) are correct. The life of the batteries appears to be about half what it was before, with my update R-D1 going through a battery in about 100 exposures instead of 200.

Another update

Having used the camera more I’m beginning to think it’s a little more complex. If you shoot quickly, then you seem to get more shots. I think there could be a constant drain now which wasn’t there before. Hence, if you shoot quickly to get a lot of shots. If you shoot more slowly the constant drain has a larger effect and you get less shots. I need to run some practical tests on this really.