Review of Haoda's split focus screen

Review of Haoda Fu's Split Image / Microprism focusing screen for the KM 5D
© Arto Rantala

To celebrate reaching 100 posts on Dyxum, I am posting my review of the Split Image focusing screen for KonicaMinolta Dynax/Maxxum 5D, supplied to me by and available from Haoda Fu. UPDATE: Haoda has a new web site at http://haodascreen.com

First impressions

The focusing screen was shipped on 12th of April and I received it on friday the 21st in a standard bubbled envelope. The envelope contained a pink plastic case reminiscent of a memory card case. In this case was a small plastic pouch that contained the actual focus screen.

The screen itself was accurately cut to the measurements I had provided Haoda with, but otherwise looked plain. There was no tab like in the original screen nor were there those little pegs that hold the original screen in alignment with regard to the AF/AE markings screen. I can't say I expected it to have any of those.

UPDATE (10th Jan. 2007): Haoda has informed me that this has been fixed and new screens come with tabs to provide proper centering.

Precisely in the center of the 1.5mm thick plastic screen I saw the two horizontally divided semicircular prisms. Encircling those is the microprism zone, consisting of dozens of tiny pyramids. The rest of the screen is a matte surface not unlike the standard 5D focusing screen.

Installation

After taking the screen out of its protective packaging, I proceeded to drop it on the bathroom floor that was fortunately covered by a soft carpet that unfortunately happened to have my camera placed right in the very spot where the screen fell. From the lens cap the screen ricocheted to hard floor that is tiled with rough-textured tiles. Stress testing completed. Damage: zero. Well, a tiny scratch. Not a problem. At least it didn't shatter.

I removed the lens from my camera and took a look in the upper part of the mirror chamber. There I could see the current focusing screen, held in place by a metallic clip. With the camera laying on its back, I pushed the handle of this metallic clip, which made a snap sound and became loose. I placed a piece of a Pec Pad on top of the mirror and let the clip rest on top of it.

Along with the clip fell out the matte focusing screen, a thin shim and the markings screen. I removed the matte screen with tweezers, noting that the matte side was facing the pentaprism. After placing it on a piece of paper, I turned the camera on its roof with the tripod socket facing up. This allowed the markings screen to fall back in place along with the metal shim and clip. I removed the loose clip and knocked the camera a bit to ensure that the mentioned parts were fully where they were supposed to be - on the bottom of the hole. You should note that the camera is not stable when placed upside down and will easily fall on its back, so keep it supported (don't ask how I know).

I then took the new focusing screen and checked which side was matte and which wasn't. It's not easy to see this at first, but by turning it so that the surface reflects a light source directly, it's obvious which side is matte (diffuse). I uneventfully dropped the new screen into the hole with the matte surface facing the pentaprism and put the metal clip back in its place. I did have to knock the camera a bit again to make the screen fall to the bottom. After making sure the clip was secure, I just attached a lens and started playing around!

Practical matters (This is no longer relevant information, see UPDATE under "First impressions" above)

The first thing I noticed with the new screen was that it wasn't perfectly aligned with the viewfinder markings. The horizontal autofocus mark directly to the left of the center mark was partially overlapping the microprism area edge whereas the AF mark on the right side was fully inside the microprism area.

I had a hunch so I took the lens off and started poking the focus screen around. Turns out the box for the focusing screen is a couple of millimeters larger than the screen itself. Since the new screen didn't have pegs to fit in the matching holes of the markings screen, it wasn't held in place properly. I tried to align it better, then engaged the clip, reattached the lens and checked the viewfinder. Much better. Unfortunately, after moving the camera around a bit it was off center again. Apparently the clip doesn't hold the screen down strongly enough to prevent it from moving. I've talked to Haoda about this and he says that he has used it on a 5D and hasn't had a problem and suggests that I may have installed it wrong. I have checked, but do not concur with that assessment. After using the camera for a while, I've noticed that the screen seems to stay in that off-center position quite reliably.

Does it work?

Screen placement is not a major problem though. So what if it's a bit off center! The main question is, how does it work for focusing? The answer is: pleasantly well. So far I haven't noticed any problems with focusing accuracy. It's right where it's supposed to be. A couple of times I have noticed some slight deviations, but I haven't been able to reproduce them reliably and since I've never done focus testing as meticulous as this before, it could just as well be because of the camera or because of me moving the focus ring or the camera itself very slightly after focusing. Essentially, for any focus errors I've seen, I am not blaming the focusing screen. Below you will find images of the focus screen in use with a scene that is out of focus and then with the same scene with the black cable in focus.




Aperture considerations

Being fairly new to split image focusing screens, I was entirely unaware of a phenomenon associated with these screens. If you do not currently know the principle of how a split image focusing screen works, I highly suggest that you read this article at least up to page 12.

To summarize, the split image screen uses light from the very edges of the aperture to achieve its effect. Due to this, the aperture used must be fairly large to actually provide such light. If the aperture is too small, there is no light refracted towards your eye and at least one of the split prisms becomes black, rendering the screen useless for focusing. The same applies to the microprism field around the main split area since it works on the same principle, only in smaller scale.

In my tests of this particular screen, I found that until around f/4 there isn't a major problem from a practical standpoint. After that, it becomes a bit too easy to begin to black out one of the semi-circles if you move your eye a bit to the left or to the right from the viewfinder centerline. After f/5.6 it doesn't seem to be possible to keep both semi-circles fully visible at the same time. From f/5.6 onwards I'd say this screen is more of a disadvantage than an advantage in focusing.
At f/2.8 it's impossible to make either semi-circle black out.

Please note that all focusing is done in the maximum aperture that the lens supports, so this issue only causes problems in situations where the lens is physically incapable of reaching larger apertures than those mentioned. It doesn't affect situations where you want to use a smaller aperture when taking the picture as the lens is only stopped down when the shutter is pressed. For example, I never see a problem with my 28-75 f/2.8, but my Tamron 70-300 f/4.0-5.6 is a bit problematic at the 300mm end, as is the kit lens at 70mm.

As you can read in the above linked document, this effect can be attenuated at the cost of focus accuracy by reducing the refraction performed by the prisms. In effect, you might find a screen somewhere that can go up to f/8 without blacking out, but it won't help that much with focusing anymore. Haoda's screen seems like a fairly decent compromise in this respect. I can't recommend it to anyone who uses the 500/8 or other such mirror tele frequently.

Brightness and Automatic Exposure

First, the brightness. Below you'll see two photographs taken through the viewfinder. In both cases the camera laid in exactly the same position, facing a white object lit by a single 500W light. Both photographs were taken with a Panasonic DMC-FZ5 using manual settings of ISO100, f/4.0 1/200s, Tungsten WB. I tried to get as much of a similar view through the viewfinder as possible.




With the standard screen, a 5x5 color sample taken from just below and to the left of the top right focus mark gives RGB values 204, 192, 139 respectively. A sample from the same spot in the image of the new focus screen gives values 186, 170, 124. Based on this, the matte portion of the split image screen is not as bright as that of the standard screen. The numbers are closer to equal in the central area, but the standard screen still seems to have a slight edge even there. Probably within the margin of error though.

Now, the automatic exposure. This is an area where I believe the most major of the hurdles lie. It is definitely affected and it would seem that different exposure areas are affected differently as well. My experiments have produced quite varying results, but I have come to the conclusion that the same issues that plague the viewfinder image also harm the AE metering. I have so far noticed that the matte area, i.e. the non-spot area, is always metered slightly wrong, because the camera is calibrated for a brighter matte screen. This means that the camera tends to overexpose these areas by about a stop because it thinks they are darker than they really are. Now, if that was the only problem, it would be easy to fix with some exposure compensation. Unfortunately there is a much bigger problem with the spot metering area.

As I discussed earlier, the microprism area, which for all intents and purposes matches the spot metering area, is significantly brighter than the matte area. This means that when the camera, equipped with a large-aperture lens, is pointed at a monochromatic uniformly lit wall, spot metering produces underexposed results when compared to full honeycomb metering. However, the blackout issue that affects the viewfinder also affects the metering. If I perform the same test with a large-aperture lens, pointed at the same wall, the spot metering in fact overexposes. This is presumably because the spot meter is seeing at least one blacked out semicircle as well as blackouts in the microprism field, thus reducing significantly the amount of light falling in the spot meter sensor, making the camera think that the scene is darker than it really is. From my testing, approximately f/3.5 is the aperture where the spotmeter reading matches the multi-segment reading.

Now if that wasn't enough, there is one more issue with the spot metering. Since the spot meter is apparently not located precisely in the center of the view, it will, at smaller apertures, always see one of the semicircles darker than the other. This means that taking a spotmeter reading from a target with the camera in a portrait orientation can give much different results with the camera turned 180 degrees. These pictures were taken using the spot meter with the camera first tilted 90 degrees left, then 90 degrees right for the other one. For both pictures, I placed the center marker right on the line between light and shadow. Please note that I did a lot more investigating than just taking these two pictures and the camera constantly metered significant differences depending on which way it was rotated.

UPDATE: I should note that I saw this asymmetrical behaviour with both, small and large aperture lenses. I originally tested this with only the 70-300@300/5.6, just to verify my theory about the blackouts. The fact that this happens even on the 50/1.7 still doesn't disprove it, only indicates that the lightmeter is so far from the central path that much less light falls on it from one of the prisms. It does however raise questions about the reasons for the spot metering over/underexposure at different apertures, which I talked about in the previous paragraph. Needs pondering...



If any of you have knowledge of the inner workings of the 5D, I would be particularly interested in knowing how the 5D's light meter is situated. This could help me confirm the reasons for this issue. Unfortunately, I doubt it's possible to fully fix the metering issues even by having the lightmeter recalibrated by Minolta service. It could possibly be mitigated though so that only smaller-aperture lenses are seriously affected.

Opinions and Recommendations

I have now had the opportunity to spend a couple of days with Haoda Fu's split image focus screen and while the screen hasn't been everything I hoped it to be, it is still far from a disappointment. I have yet to decide for myself whether I will keep this screen or return to the standard one. On one hand, it is most helpful when using fast lenses for standstill subjects at short distances. It really does make focusing easier in such situations and if while I lack decent close-up equipment I'd imagine it would be even more useful with macro. On the other hand, it is not helpful at all when shooting tele with a slower zoom lens. The metering problems, blackouts and the simple fact that the screen isn't really helpful with moving targets with few vertical lines such as birds make the entire experience more frustrating than it was with the matte screen. Perhaps getting more used to the screen would help in the long run, but right now telephoto is not fun at all. I found that the AF confirm dot usually informed me of proper focus much before I could confirm it myself from the actual viewfinder.

Where most users would find this screen most useful is, I'd imagine, vintage manual focus lenses such as those with an M42 or Canon FD mount. Unfortunately I have none of those yet. If I understood correctly, the focus confirmation dot is not available with these lenses so you'll really need a good way to confirm the accuracy. I must warn that focusing may not be faster with this screen, but it will help make it more accurate so that you won't need to check from the LCD afterwards to see if that shot was correctly focused or not. Unlike with film though, such accuracy is rarely mandatory with digital because you can check your shots easily and just shoot again if you missed it.

My recommendations after a couple of days of use and testing are these:
- If you use M42 lenses, if most or all of your glass is faster than f/4 and you don't do much telephoto, do get this screen. While it may not be everything you think you've always needed, you'll still find it useful and you should be able to anticipate and compensate for most exposure problems.
- If you have and use slower lenses a lot, depend on accurate metering and have a good enough eye to do decent manual focusing without this screen, don't get one. You'll find it's more trouble than it's worth.
- For those of you like me who enjoy carefree metering, mostly use fast lenses, can't afford those f/2.8 teles, but like to pop off a shot or two every now and then with your 70-300, this can be a real dilemma. Right now, I think I'll probably go back to the usual matte screen, but will keep the split screen around for when I get my paws on some M42 optics. I'll try it again then and see if it's worth keeping. I really do like the the help it gives in some situations, plus it has quite a bit novelty value. I just don't know if that's worth all the trouble.


This review is Copyright © 2006 Arto Rantala.
Please contact me for permission if you want to post it elsewhere.

Edit: Fixed some typos and changed some sentence structures. I apologise for those I didn't change. =)

Edited by Gubbe - 10 January 2007 at 10:51

Congratulations on an excellent review.

Congratulations on marking your 100th post with this review. Well done Arto!

I guess your findings are some of the reasons why the designers dropped the split-screen feature. Great observation on noting how a semi-circle blacked out when you moved your eye off-center of the viewfinder, it portended your findings in problematic metering. What's really interesting is the 90-degree switch-yikes! But that makes sense since your contrast line changes from a horizontal to vertical orientation. While a grey card would have provided a better lab test, this real-world exposure was much more helpful. Sounds like 1 semi-circle is weighted more than the other (and it may indicate where the truer metering position is.) The darkening at smaller aperatures was also problematic when using the DOF preview button on my XE-5.

My thoughts from your findings: Barring a change to metering components and position, a remaining variable points to the penta-prism. The 7D uses a chunk of optical glass achieving it's "mirror" by polishing, the 5D pentaprism achieves it's mirror by an applied coating. Is a mirror, a mirror or are there subtleties in these two mirror types that alter light behavior and metering preformance? Hopefully, a 7D user will conduct a similar test.

Once, again, well done and THANK YOU!

Once again i want to express my gratitude to Gubbe for both , this excelent review and the permission to host it on-line (we have managed all details and the review is up now)

Hi all

I have ordered two of those screens for my 7D.
They work great, but I need only one.
Drop me an email if you're interested.

Thanks for a great review!
I won´t buy one, too expensive and light metering issues...

I'd very much imagine that the alignment could be fixed with some padding around the edges. I haven't spent a lot of time considering the precise specifics of how to do it, but adhesive tape would probably be a good starting point.
Just be careful and make sure that the tape stays where you put it. Strips of tape might not be the most pleasant thing you can discover on your shutter curtain...