Alpha 900 – finder and frames
It’s not going to be long before we see the Alpha 900, and some cameras are known to be out there on trial in the hands of Sony staff and pre-release testers. I am not one, so rest assured, this is not a leak! What can you expect from the Alpha 900’s full-frame prism finder?
(Note: this post was written in early August – it is now 100% certain that the finder is 100%, and at 0.74X magnification will be – as had been hinted – the largest of all current DSLR finders in apparent visual terms except the EOS 1Ds Mk III which is 0.76X. Comparisons: EOS 5D 0.68X, Nikon D3 0.70X).
Well, I’m going to risk being shot down and say that I think it will be 95% coverage, like the Nikon D700, but could even be slightly less than 95%. Why?
The Alpha 900 has full-frame SSS, in-body image stabilisation. Photographs of the inside mechanisms indicate that it does have an oversize ‘film gate’ and shutter assembly, to allow for the movement of the 24 x 36mm sensor. There is some possibility that optical vignetting may occur when strong SSS correction happens during an exposure, but I would hope no mechanical vignetting is allowed.
Optical vignetting happens if the sensor moves to the edge of the lens image circle. Mechanical happens if the sensor moves beyond the rectangular frame created by the shutter opening.
Why should this affect the viewfinder? If the finder was a 100% view, it would only be accurate when the sensor was parked – SSS turned off. It’s difficult enough to make a 100% viewfinder match up perfectly to a static sensor, as Canon and Nikon have both found. SSS would have to park the sensor assembly dead centre to make it match a 100% viewfinder.
Therefore, a 95% view through the large prism seems far more likely. 95% is not a big trim round the edges of the field of view, but it’s enough to ensure that the sensor more than covers everything when SSS is turned off. It also makes it less likely that a displacement of the sensor, with SSS on or not precisely parked, could produce a composition significantly different from the viewfinder view.
A carefully composed Alpha 700 shot with AS enabled, showing the theoretical 95 per cent viewfinder coverage and the actual frame. The subjects are offset to the left in the image, but were centered in the finder; however, all of what could be seen through the finder is present in the overall image (the crop shown would be moved to the left to restore the composition).
Of course the Alpha 700 is a 95% finder. In theory it is possible to have SSS shift the sensor so far, at the moment of exposure, that you might fail to get exactly what you saw through the finder on your final shot. (See comments to this post – ‘in theory’ you should always get what you composed! – I have now run tests and the variations in sensor position or parking are very small, between 10 and 20 pixels, with just the occasional shift big enough to be noticed).
In practice, the movement of the SSS mechanism is tiny – measured in a few pixels at the most, and as each pixel is only around 1/100th-200th of a millimetre, displacements of 2.5mm or 5mm are very rare. So on the whole, we get in our image what we composed in the finder. The 95% view provides a good margin of safety.
Would Sony dare put a 100% viewfinder in a camera with SSS? It’s just possible they might do so. They might print a small warning in the user manual to state that for accurate use of the 100% viewfinder, SSS should be turned off. But a 95% finder would be safer for the user and Sony alike.
(Update Aug 22 – several sources insist that the Alpha 900 finder is not only the largest and brightest of any DSLR now on the market, but definitely 100%. My view is that this is good for marketing, but we can expect some reports of sensor misalignment; those have happened with all the KM/Sony DSLRs so far, and every other make I know of except Olympus.)
Windowed readout
This section has been amended – the Alpha 700 doesn’t window 16:9 ratio files. The final release of ACR 4.5 treats 16:9 raws from 700, 200, 300, and 350 identically – locking out the full raw file size which is actualy present. The earlier beta release cropped A700 files but did not crop the latest model files.
The Alpha 900 sensor is said to feature ‘windowed’ readout, a function also found on the Nikon D2X and D700, and on the Nikon D2X models. Windowed readout allows different crops or aspect ratios to be selected. The Alpha 350 does not feature it for the 1.5X and 2X crops when in Live View mode, but the 16:9 mode results in a visually cropped raw file in correctly aware raw processors (as on the 700/200/300).
What is the difference? If you own an Alpha later than the 100, you can try it out. If you shoot RAW and set 16:9 image ratio (HDTV format) you will discover that not only your JPEG is cropped – your .ARW file appears as cropped too. But it’s not. Examination using a non-aware raw converter like Raw Photo Processor or Capture One Pro shows the full 3:2 raw. ACR 4.5 and Sony IDC show the cropped area only. Sony Image Data Lightbox shows the cropped thumbnail, but if you double click to preview, reveals the rest of the image slightly greyed out top and bottom.
In theory it could improve continuous sequence shooting rates in a model where the raw file was genuinely cropped. In the Alpha 350/300, the sensor readout is NOT ‘windowed’ to get the 1.5X and 2X ‘Smart Teleconverter’ crops – a full size raw file is captured and processed (whether saved or not) to create whatever resolution and digital crop you select, and there is never any boost to sequence shooting rate. And you can’t use the Smart Teleconverter in raw-only shooting mode, as it only affects in-camera JPEGs.
A 16:9 raw file next to a full frame raw file, from the Alpha 700
The Alpha 900 will feature at the very least a 16:9 crop and a 1.5X crop, to allow the use of DT (APS-C format) lenses such as the Carl Zeiss 16-80mm with a 10 megapixel file size. The viewfinder will have frame markings for the 1.5X crop, and the usual top and bottom lines to indicate 16:9. I’d like to hope it will have a 2X mode, a still-useful 6 megapixel image. The 16:9 and 1.5X are not optional choices for Sony. They must include these and the sensor is equipped to provide them. Anything else, like 2X or 5:4 ratios would be optional.
Windowed readout can create cropped RAW as on the Nikon models, so if Sony provides the option to use it with full frame lenses – again I do not know whether they have done so – caution would be needed in using the finder. You get genuinely cropped raws.
The benefit is possibly blistering motordrive speeds. Assuming the Alpha 900 has been configured to manage at least 3fps at full 24 megapixel resolution when shooting raw, it should be able to achieve 7fps at windowed 1.5X 10 megapixel resolution. We might well see 5fps full frame and better than 10fps APS-C – now that would put the Alpha 900 in an interesting market position!
It’s all down the how the raw saving is implemented. If it is genuinely cropped – for 1.5X if not for 16:9 – faster sequences may be on the menu. If the raw file is always saved at full size, and just includes metadata cropping information, it may be possible to use DT lenses like the 11-18mm and get a useful full-frame wide angle at 14mm with a forced crop.
Because the Alpha 900 is designed to use APS-C lenses as well as full frame, it will be interesting to see what area the AF sensors cover. Will they be clustered centrally within the 1.5X image zone, or scaled up to work optimally with full frame movement tracking?
Electronic shuttering
The final possibility offered by the new sensor is an electronic shutter. I think there is a 50% chance the Alpha 900 will have one. Why?
The 900 has the largest focal plane shutter assembly of any 35mm-format camera (DSLR or otherwise). I don’t know the exact gate size, but it must be around 40 x 30mm. So far, even with a fast 1/8,000th shutter fitted to the Alpha 700, Sony has not achieved particularly fast flash synch speeds. The A700 can be used safely in the studio at 1/250th with zero vignetting but the autoset speed with the dedicated flash system is 1/60th-1/200th (in most conditions it sets 1/125th or 1/160th).
Alpha 700, full frame in the studio - no cutoff at all with studio flash cord sync at 1/200th. This image was centered up just as carefully as the first one and is very accurate relative to the viewfinder composition.
The extra size, and transit opening, of the Alpha 900 shutter gate could mean a sync speed slower than 1/160th unless the shutter itself has been speeded up to a 1/12,000th specification – something very difficult to get in a unit bigger than that used for the 9xi and Dynax 9. My gut feeling is that to avoid reduced shutter life and a loud shutter sound Sony will restrict the Alpha 900 to 1/4,000th like the Dynax 7D (a camera it closely resembles in many ways, including the return of the 1-2-3 memory registers).
This would mean a flash sync speed no better than 1/90th with SSS active, perhaps 1/125th for PC sync connections or non-SSS, non-wireless operation.
The provision of electronic shutter mode on the full-frame sensor would make it possible for Sony to offer faster ‘virtual’ flash sync speeds. The Nikon D70 did this, enabling synchro sun at 1/500th. Flash timing latency, inevitable in an electronically triggered system, may prevent anything faster. Having options of 1/250th or 1/500th would make the Alpha 900 appeal to fashion, portrait, press-PR and wedding photographers as all regularly use synchro-sun fill in.
(Update August 22 – it seems that Sony has taken the first option – though the highest shutter speed is 1/8,000th like the Alpha 700 and the ‘gate’ is much larger, they’ve achieved a faster flash sync speed than the Alpha 700 without electronic shuttering- there’s no question of it being restricted to slower speeds. What this means for shutter noise/life will depend on the construction of the shutter. I think we are going to be given details of a very advanced shutter design using lightweight materials. How can they do this, and keep the price down to the low figure hoped for by so many? – DK).
If Sony has implemented electronic shuttering, the ultimate bonus would be burst mode ‘open once, capture sequence’ shooting – where the shutter opens for a short time and a sequence is recorded at a higher than normal frame rate, perhaps with reduced resolution. This is a CMOS technology Sony has been working on for the last few years.
(Further update – I do not think, if my latest information about the focal plane shutter is correct, that the electronic shutter with very fast burst shooting will appear in the Alpha 900. I am certain that it will appear in a future Alpha DSLR. – DK)
You’ll have to wait until mid-September to learn whether the Alpha 900 will be more than just a moderately priced full frame DSLR. I don’t think you will be disappointed.
– DK
Thanks David. When I’m using flash I’m usually in manual mode, selecting the appropriate ambient light exposure for the conditions. The exception to this is when shooting fill flash outdoors in “decent” light, where I know there is sufficient ambient light to get a decent ambient exposure with the camera in A mode, and where the camera is more willing to set a shutter speed higher than 1/160.
As an aside, what I wish the A900 would have for flash shooting is a “balanced fill” mode, where the camera will set an ambient exposure that won’t result in dark (black) backgrounds as frequently occurs when shooting flash in darker interiors when the camera generally defaults to 1/60, which is not sufficient in such conditions to result in an ambient exposure. I have actually discussed this with a Sony rep at the big New York fall photo show (someone from Japan who apparently has some involvement with product development). This is something that would help wedding and event shooters who need or want some level of ambient exposure with their flash. Ideally, such a feature would allow the user to define the slowest shutter speed (in A mode) that the camera could set to avoid getting too slow a shutter speed (which you could still do or use in manual mode). In essence, it would be the same as using the AEL button (“slow sync”) when shooting with flash, except that the meter would remain active so you don’t have to continually press the AEL button to get the “slow sync” results in changing light conditions (indeed, it could be an option for the AEL button and the toggle function). I believe that Nikon offers a feature like this in its higher end cameras (though I’m not 100% certain). Were Sony to add this capability it would be a positive sign that it is learning a bit more about how photographers actually shoot and what features or capabilities are truly useful.
Thanks again for all your good work.
Mark
Quite right. Of course it sets 1/160th nearly all the time – or slower – so you get used to the idea that the X sync is still 1/160th. The Alpha 900 will match the best for X sync I’m sure (someone who has had one in hand for a few minutes apparently managed to suss out or be told the maximum X-sync – I think they may have been told, as they did not have a flash gun and I don’t see how you can find out otherwise. ‘Told’ may mean anything from accurate information to wishful thinking even from a Sony employee, so I will not repeat the details. Let’s just say it is unexpected – I was pretty sure it would have to be slower than the A700). I’ll amend the article.
David
David,
Thanks for your thoughts and follow up comments. However, I noted an error in your discussion of the flash sync speed for the A700. It will sync at up to 1/200 with SSS active, 1/250 with SSS off. The 7D had the slower sync speed you mention, 1/125 with SSS and 1/160 without (if I recall correctly). The general discussion of flash sync speed for the A900 is important, for the reasons you noted, particularly if Sony has any thoughts of trying to market this camera to wedding pros (and it should given the fast CZ and G glass available, fixed and zoom, combined with SSS, which gives it an edge on the competition for available light, or low light shooting in many situations. Of course, good high ISO quality is another important factor there, and it remains to be seen what that will be like. 🙂
On the subject of windowed readout, since the width of the APS-C frame is very close to the height of the FF, I wonder if SONY has thought about providing a vertical cropped frame for DT lenses (or FF lenses). This will enable people shooting vertical frame without having to rotate the camera.
I have revised the article to change the 95 per cent view graphic (1st image). First of all, the original under-represented the 95 per cent margin. My calculations for viewfinder apparent size, as published in the British Journal of Photography (and copied by DigitalRev’s website!) are based on Nikon’s statement that their viewfinders show ’95 per cent (horizontal and vertical)’ which clearly indicates the percent is given for the linear dimensions. Some websites say ‘95% of the image area’ but this is misleading. A 95% coverage (horizontal and vertical) finder shows 90.25% of the image area. Secondly, my shifted frame indication was misleadingly extreme; the actual positioning error of SSS would be within the margin between the 95% and 100% zone, on the A700, the biggest displacement I could get was 3.5% linear off-centre.
David
Ref ET: “I hope they do allow a DT lens in full frame mode.” I can’t confirm this. I do know that the Alpha 900 will have a cropped mode for DT lenses, but I do not know whether this cropped mode is also available for full-frame lenses (I assume it is); or whether fitting a DT lens will force the crop mode, and lock out full frame.
On dPreview there is a thread about possibly using the 18-250mm etc. No-one has replied sensibly yet – the 18-250mm is no good for full frame. At no focal length do you get a clear frame. The 16-105mm produces a surprisingly large 16mm circle which would crop to a clean square given some PTLens type corretion, but only at exactly 105mm does it almost make a decent full area. The 16-80mm never does, and its 16mm circle is no better than the 16-105mm. The 11-18mm is the real winner on full frame, as it can be used at 14mm and stopped down to f11 for an almost perfect result.
So yes, it would be great if the APS-C/DT function of the Alpha 900 was either not automatic, or could be over-ridden when a DT lens is mounted.
David
Here is the bad news: I have now done the test without swinging the camera round. Just turning the camera on its side, either way, tilting it for an angled shot, holding it upside down – all result in the sensor not parking in exactly the same position. While it is very consistent with a normal horizontal composition, a different orientation results in a small change in the sensor position. This is occupying too much of my time when I’m supposed to be working, and I know I should now test all my other Alpha bodies to see how well the sensor parks.
But, from my few tests on the Alpha 700 (the latest one done with a remote release to avoid any possibility somehow stressing the assembly) I can see that the sensor does not locate with 100 per accuracy, therefore a 100 per cent viewfinder (at least on the Alpha 700 mechanism) would be prone to composition errors.
David
David, Thanks for doing the test. I agree if the SSS is functioning correctly, the background should be sharp while the target blurred. How about some more modification of the test: try a 18-250 at the long end to make the SSS work harder, use high shutter speed to minimizing the motion blur, and instead of swing the setup, rock it at a high frequency. (leave it on the hood of a parked car with engine running?)
I hope they do allow a DT lens in full frame mode. With the CZ 16-80 you could crop a nice square format post processing – that would result in 16 megapixel resolution. Though with the CZ 16-80 you would get the familiar vignetting across more of the frame if used wide open – I tried it with my 600 si film camera. When I tried the Tamron SP 11-18mm F/4.5-5.6 Di-II it seemed I was getting full frame coverage from about 15mm. We’ll need to wait and see what sharpness across the frame will be like, but stopped down to f11, it might be an interesting option.
I’ve now done a better test. I made a construction using Kennett brackets, to hold a target in front of the lens, and shot at 5 fps while swinging the Alpha 700 around (not shaking it, just changing movement direction). The sensor does not repark dead centre and frames show different positions for the target, including the very edge of the assembly which is unlikely to ‘move’ due to flexibility. the greatest deviation is significant, but most frames are very well centered.
I repeated the test using single frames, setting the swing in motion before pressing the shutter, and the results are similar – the A700 sensor may have moved between shutter press and actual exposure.
But, I think the test is flawed. While it does detect a shift in sensor position at the moment of exposure, in real situations the subject is not attached to the camera, and the degree of movement I’m using far exceeds typical tremor shake – you’d have to be using the camera on a fairground ride to get the same forces.
Also, SSS appears to be disabled or nonfunctional under these conditions (similar to fast panning with action). The fixed target should be showing a motion blur equal to the movement of the sensor during exposure, but it doesn’t. Sony has said that the SSS system is compatible with panning, meaning that after the initial acceleration the sensor would presumably be static but not centered (think about it – it would a disaster if it jerked back to centre during a pan having made a small deflection on first detecting movement).
David
I have just done your excellently devised test, using a tripod. It remains very stable, but one in maybe every dozen frames (perhaps because of the direction I shook the assembly) shows a slight shift in position. At 16mm aimed at the tripod leg, it wasn’t easy to spot any shift. At 80mm, some displacement could be seen. Whether this is due to slightly misplaced parking, or an effect which can happen any time, I’m not sure. I will repeat the test using a fixed focal length (more stable than the 16-80mm zoom which might even be able to shift the image itself under G-force) and outdoor light, and perhaps also using single frames.
What’s a bit odd is that the tripod leg stays sharp and the background can be a total movement blur, with SSS on. Surely the SSS should be trying to cure the shake, which should produce a blurred tripod leg? Yet in most of my tests the tripod leg (locked in position relative to the camera) was sharp and the background shake-blurred.
David
To verify if the sensor is always reset to the center location before every shutter opening, one can try the following experiment:
Attach an A700 and a target firmly on to a rigid object such that the target is focused and fixed in position in the viewfinder of the camera (e.g. camera on a tripod with lens pointing at one of its legs). Now shake this assembly while firing the camera in high speed continuous mode. If the camera resets the sensor before every shot, the target object will be at the same location in all the images. If the target changes its location between frames, then the sensor is not always reset.
Very thoughtful article. I look forward very much to seeing the A900 specification. I have been wondering about the target buyers Sony have in mind for this FF camera. Hasselblad & other medium format Digital camera makers are moving up to 50 Mp backs. A FF DSLR at 24 Mp should be a considerable threat to medium backs up to 22 Mp if it can produce low noise images up to ISO800 – particularly if it can operate teathered and has sufficient CF card storage and instant transfer ability to laptop for plenty of images as well as quality lenses – prime & zoom. I really do think it will have to out-perform the EOS-1Ds MK III in terms of image quality and noise at least up to ISO 800 to have a real chance to break through in numbers sold. I think that David is also right in highlighting the need for a high flash sync. speed to attract a segment of users where this aspect is important. I do not think I will be needing one as my A700 & A350 can cover my requirements as they are just now. One benefit could be that the A900 is helping to drive Sony to release more quality lenses and also to force Sigma/Tamron to make more of their quality line-up available with the Alpha mount. As any new FF lens can be used on the A700/A350 that may benefit me. A final comment that has not been mentioned in the exitement over the “hardware”. If Sony do not release a V. 3 of their IDC that matches what is availble from Canon & Nikon and incorporate Chromatic Aberration removal I think they will shoot themselves in both feet in one go. I was really disappointed when I saw V. 2 and its lack of CA removal considering the ability of the CZ 16-80 to produce plenty of this nuisance. I need to check every picture taken with this lens for CA. Maybe mine is particularly bad.
I’m wondering why they’re cramming 24 megapixels into it, myself. I do a lot of high ISO, low light work and I’m really concerned about pixel density.
I hope they can give it a 100% viewfinder, that would be very nice. Interesting times ahead in the dSLR market, that’s for sure.
Interesting article. I also think 100% viewfinder may not be possible if Sony plans to add liveview with SSS on the body. I think any value from 95% to 100% is possible.
I think Sony is playing big and will try to surprise us in many ways. Except the 24 MP, IMHO other items should also differ from previous Alphas.
First some thought and insisted that FF with SSS is impossible which (the idea) was soon “dethroned”. Now we’re having doubts that SSS with 100% viewfinder is not possible… Maybe it is?
Anyway this article is very interesting and as TallPaul said “we will find out soon…”
Point taken – in theory you should get what you composed. In practice I’ve found otherwise, especially with action shots (pan and fire) and sequence shooting (maybe the sensor does not re-park between each frame). Also, depending on the exact moment the sensor movement begins, there is scope for tremor-type shake (around 10-20Hz) to go two ways before the actual exposure.
I think my diagram shows far too much displacement anyway, the real movement would be within the 95 per cent to 100 per cent zone crop. I’ll remove the reference to Canon as the issue is probably one they considered and decided that in theory they might be wrong 🙂
David
I think you have the sensor movemnt description somewhat backward. If my understanding is correct, the sensor stays centered until the shutter fires (or possibly the mirror flips) at which point it tracks the movement of the camera (i.e. moves in the opposite direction to keep the image stable). If the tracking starts with the mirror flip, then a 100% finder would give you exactly the image you saw at the instant you pressed the shutter. If it starts with the actual shutter, it won’t be off much. Either way, you can’t do much better to match the finder image. I have never encountered lens vignetting with my A100 and I have taken some pretty shaky pictures with long exposures, so I doubt if that will be a problem either in anything but an extreme case where vignetting will be the least of the problems with the picture.
Interesting theories, one things for sure… we will find out soon we hope!
Good read, thank you David.