BMPCC + Speed Booster = shifted focal plane?

[sir_danish]

Hi,

I´m using a Sigma 18-35 1.8 with a Metabones Speed Booster on my BMPCC.
What I noticed is that the left side of the image isn´t perfectly in focus while the right side is spot-on.
If I try to get the left side as sharp as possible, the middle of the picture and the right side will be slightly out of focus.

Please take a look at the pictures below. Notice how the very left side seems to be quite blurry, while the right side isn´t that bad. The aperture was 2.5.
'>
'>

I´ve tried two Sigma 18-35 and two different Speed Boosters, but the results were exactly the same.

Has anyone else noticed similar effects?

 

I would be very happy if someone with a Sigma 18-35 and a SB tried to shoot something outside with the lens wide open and focus set to infinity.

 

 

[Henry Gentles]

Your shooting through a window, you need to go outside and shoot a brickwall?

[Nikkor]

Have you tried the lens on another camera? It might be broken.

[sir_danish]

The window has no influence on this effect, which shows up outside as well.

I´ve done several tests and posted the results here:

http://forum.blackmagicdesign.com/viewtopic.php?f=2&t=21564

 

Because nobody on the blackmagic forum with the same camera and lens wants to test it, so that I could compare the results,

I´m aksing EOSHD members for help.

 

It´s a fact that it´s not possible for me to get the left and the right side in focus simultaneously, without misalignment of the center.

I doubt that it has something to do with the lens, as it works fine on my GH3.

 

Because I have no second BMPCC to test, it would be very helpful to see if other BMPCC+Sigma 18-35+SB owners are getting the same results.

 

Thanks in advance!

[Henry Gentles]

I'll try it tomorrow morining, but I only have an Eos to M43 Adapter, I have tried my 18 -35mm Sigma on my BMPCC but haven't recorded anything to sd card yet with it. I dont have a speedbooster, my sigma is canon, not sure if it will help you though? 

[sir_danish]

I'll try it tomorrow morining, but I only have an Eos to M43 Adapter, I have tried my 18 -35mm Sigma on my BMPCC but haven't recorded anything to sd card yet with it. I dont have a speedbooster, my sigma is canon, not sure if it will help you though? 

 

Thank you, I appreciate it! Actually, it has to be the Sigma + Speed Booster, as it could be that the SB causes the "focal plane skew".

The Sigma works fine with a Nikon F/MFT adapter on my GH3, that´s why I suspect either the SB or a misaligned lens mount on my BMPCC to be the problem.

 

The rear part of the SB is adjustable and you have to tighten it with a small screw. Maybe this screw unevenly lifts the rear lens a bit, which leads to the issue mentioned above. As I´ve tried two Speed Boosters and two Sigma 18-35, I cannot rule out the possibility that this is a common issue when using a BMPCC + SB + Sigma 18-35 1.8.

[Axel]

I´ve tried two Sigma 18-35 and two different Speed Boosters, but the results were exactly the same.

Has anyone else noticed similar effects?

 

I would be very happy if someone with a Sigma 18-35 and a SB tried to shoot something outside with the lens wide open and focus set to infinity.

 

Well, yes, this was to be expected, wasn't it? I have been dealing with focus compromises most of my professional life. Unless you pay thousands of dollars, you have to live with visible tolerances, but tolerances (which in a perfect world would get infinitely close to zero) are everywhere.

 

First, when I saw your images, I thought, fock, I didn't see this yet, if it's there with my Sigma & SB too, this may be the last straw.

 

Of course I made the test. It wasn't easy, because it's a sunny day, and not many things in the city are far enough away as to be focussed on with ∞.  And also, the sky was so bright, I had to darken the (Tiffen) ND-fader so much that I got a vignette at open aperture. I show you not the worst frame, but the best under the circumstances:

Vignetting, washed-out colors, and the power pole as well as the gas plant in the background look out of focus, whereas the trees on the left appear to be relatively sharp.

 

Maybe my "tolerance" is positioned on the right side  :wacko:

 

Of course it is not wise to shoot this wide shot that way. Most lenses have their best performance round about f4 - f5.6. If the goal was to capture a lot of fine detail - motif detail, not texture detail ! - I better close the aperture at least two stops. Done:

 

There is an improvement, but what you could see if the image moved a little was another nail in the coffin: *Moiré*

The white clinker brick wall on the left - see my corresponding thread, a no-go for me.

 

As I wrote there, I considered the Metabones plus Sigma the ultimate Pocket lens set up, allured by Andrews >article.

 

What it turned out to be good for: Low light shots of interiors. Full stop. It isn't an allrounder set-up, as our experiences show.

 

> It's not good for panorama shots, and particularly not for low light or high key panorama shots!

> It's not good for any long shots, because it obviously has a serious backfocus-problem.

> It's not at all good for detail (though the Sigma is sharp), rather for playing with sDoF.

 

What part is to blame? 

 

Additionally to the Speedbooster, I have a Novoflex-adapter for Nikon. I put this on with the Sigma. First thing I noticed: Whole thing suddenly seems to have more (as they say in STNG) "structural integrity". I shot stopped down about one-third (the Novoflex has no markers):

 

What happens? I don't see moire (BTW: Don't compare the colors, that was hastily done in ACR. Speedbooster was close to sunset though, Novoflex 'blue hour').

 

Next: Novoflex wide open, zoomed in. Note, how there actually is a horizon, not just a watercolored background ...

[sir_danish]

Thank you very much Axel, I really appreciate it!! I was hoping for a helpful post like this one. Is it possible to share these images in full resolution? I would like to "analyze" them, but they are too small to compare them with some of my own shots.

 

Thanks in advance!

[Axel]

Thank you very much Axel, I really appreciate it!! I was hoping for a helpful post like this one. Is it possible to share these images in full resolution? I would like to "analyze" them, but they are too small to compare them with some of my own shots.

 

Thanks in advance!

 

I tried a more sigificant test, on a tripod, with moderate light conditions. You can download the three DNGs here.

 

First is the Metabones with the aperture wide open, exposure controlled with zebra and ETTR:

 

The three points of interest enlarged:

Hard to tell, if indeed the left edge is subtly out of focus, one needed a giant test chart ;-)

 

But according to the mark on the Sigma, it was indeed infinity:

 

In comparison roughly the same framing (by zooming out) with the Novoflex, first wide open:

 

... and stopped down about one-third (meaning approx. 2-3 stops?):

 

All these images may not look pleasing, the monitor isn't properly calibrated. Also, from the low resolution (I mean the 1920 sized DNGs) it's hard to nail down if there are any differences in detail, but spontaneously the Novoflex looks clearer, perhaps less color fringing. It's no fun to watch a panorama on the muddy, peaking-contaminated display of the Pocket. Sigh, this is all such a hassle ...

[sir_danish]

Thank you very much, Axel! Your image looks quite sharp from one side to the other, with and without the Metabones SB. With my setup, the same image would be very soft on the left side. Thanks to your test, I don´t think the SB is the problem in my case, as I´ve tried two of them.

If it was a common issue with the SB, your image would look strange too, but it doesn´t.

[Axel]

Some thoughts on the Sigma/Speedbooster evaluation:

 

I am performing test shots with the Pocket since almost 5 months now. Many say the Pocket was good at low light. The cameras I worked with over the decades were: Bolex H16 Reflex (owned), Sony VX 1000 (owned), Sony VX 2000 (owned, first shots lit by candlelight), Panasonic DVX 100 (borrowed often), Sony FX-1 (borrowed occasionally), Canon XH-A1 (owned), Sony EX-3 (borrowed occasionally), Canon 7D (owned, among others adapted Nikon glasses: 35mm f1.4 and 50mm f1.4), Canon 5D MII (borrowed often, fastest lens Canon 50mm f1.8), Panasonic GH2 (owned, fastest lens Nokton 25mm f0.95), Canon 5D MIII (played with for an afternoon), Panasonic G6 (owned).

 

Now in theory the Pocket with Speedbooster and Sigma should equal - at least - the Lumix with Nokton f0.95, and more or less, it does. Nevertheless, one doesn't shoot with a 'cinema camera' to live with noisy footage that actually reminds me of high speed 16mm stock from the last century. You'd de-noise anyway, wouldn't you? Take this night clip, shot with Olympus 12mm f2.0 (more than 2 stops slower than f1.8 speedboosted):

 

No plastic feeling, well done. Two other things strike me: There are a lot of moire-traps in this clip, but it doesn't show. Would I have tried this with Sigma/SB in raw, I guess I would have had to throw away two of three shots. The other point is the absolute depth of field, something I consider desirable for Pocket shots. I have not yet figured out how to get there with our fat and heavy monster. 

 

Or this clip, starting with f2.5, almost three stops behind our combo:

(later on a good comparison with contemporary EVIL-cams on higher ISOs)

 

I conclude that Sigma & SB are suitable for filming in confined spaces where you can't control the light properly. They saved me when I ventured to shoot a portrait of a local hospice with really terrible light conditions. But can this be a task for the Pocket? Surely not.

 

Otherwise, I feel the size and weight of the glasses - let alone the price - don't fit. Bought new, the price equals that of two more appropriate lenses (if you manage to find used ones on ebay): SLR magic 12mm f1.6 and Nokton 25mm f0.95. 

 

I think I abandon the plan to put another glass (with Tiffen IR-cut and Tiffen vari-ND, the latter should obviously be substituted by a set of fixed NDs) like a diffusor in front of the Sigma. This will only degrade the quality further. Now my question is if I should keep them or sell them. Comment.

[sir_danish]

Danke Axel! Another observation: As mentioned above, when the center is in focus, the left side is out of focus, while the right side is only slighly softer than the center. Let´s presume the Sigma or the SB has a badly cut or misaligned lens. Wouldn´t the out of focus left side move to the top of the image, if I turn the Speed Booster (including the lens) clockwise by approximately 65° - while it´s still attached to the camera - just before that point where the SB comes off? I hope you understand what I mean...

 

Anyway, if I try this, in both cases the image looks almost exactly the same, despite the 65° difference. Does this point to the fact that the lens mount on the camera is slightly skewed? :huh:

[Axel]

Must be a backfocus issue, but not one related to the z-axis alone. Fractions of milimeters make a big difference. The projection-lenses in a cinema, for example, enlarge an image the size of a stamp to a 50 feet wide screen. Easy to imagine, if there is only a slight shift in the structure that holds the lens, a very similar effect as the one you describe results. What is easier in a projector: You have big screws you can turn with your fingers, you have a test chart projected whilst performing the calibration, and still it needs patience, experience and fine motor skills ....

 

You've seen this from the Tokina thread?

I believe that you actually need a rail support that very subtly allows to adjust the micro-position of the whole structure (applying a little pressure) and fixes it once and for all. Sounds cumbersome, is cumbersome.

 

Or: You just never use Sigma + SB for panorama shots. These niceties are completely lost in a shallow-DoF shot. 

[sir_danish]

Easy to imagine, if there is only a slight shift in the structure that holds the lens, a very similar effect as the one you describe results.

 

That´s why it looks like the lens mount was misaligned, as there is no difference in the image if I rotate the lens + SB. In my understanding, if the lenses were to blame, the whole image would look different after a (roughly estimated) 65° rotation on the z axis. Maybe I should get a second BMPCC for a quick test.

[Axel]

Do this: Mount the camera on a heavy tripod. Tighten all screws. Focus a brick wall or st. like that on infinity. Hit record. Press gently but firmly against the Sigma, in different directions. Then watch the clip on the monitor. I bet the oof-areas will move around the frame. There are two bayonet connectors in this construction, and they don't make a solid unit, tolerances add. Ever owned a Letus 35mm adapter? Same song. Finding backfocus was crucial. The thing was 'milled from one massive aluminum block', and they always had to be supported with rails.

 

BTW: Where do you mount your tripod? It must be the Metabones' connector. And handheld: You have to carry the weight mainly by holding the Sigma, not the Pocket. It's not so much because the mount might break, but the heavier part will bow.

EDIT: An elegant way to stabilize the structure, instead of rails, could be by using a tripod plate. This was my first makeshift 'grip' for the Pocket:

 

You could press a piece of rubber or so between the ground plate and the non-moving ring (where I have my thumb in the picture). One can at the same time 'fix' the position and hold the camera on the lens with both hands:

[sir_danish]

I do stabilize my camera and lens. The camera is mounted on the pocket rig from edelkrone, the Sigma rests on a lens support.

http://i1285.photobucket.com/albums/a588/sir_danish/20140326_172322_zps1c2bbf67.jpg~original

 

You are absolutely right, the out of focus areas move around if I grab the lens and put pressure onto it, because the whole construction is a bit loose. The lens has a tiny bit of play and the SB does too. But that´s what I got the lens support for and even though the lens should be almost perfectly aligned, there is this out of focus left side. My first Speed booster didn´t have any play at all, yet I had the same problem with the out of focus left side. Exchanging the SB and the Sigma didn´t show signs of improvement.

 

This is how it looks like if I put the camera in front of a sceen with a grid pattern:

http://i1285.photobucket.com/albums/a588/sir_danish/18mm_zpsb108c76a.jpg~original

http://i1285.photobucket.com/albums/a588/sir_danish/35mm_zps2359757a.jpg~original

 

Now take a look at the focus peaking. Notice how the in focus center seems to be shifted towards the right side, even though I did my best to align the lens and the SB with the lens support.

http://i1285.photobucket.com/albums/a588/sir_danish/focuspeaking_zpsbc76972d.jpg~original

 

I know, a wide shot with a fully opened lens isn´t the best thing you could do, but I´ve already been in low light situations where this was absolutely necessary. Everything looked fine to me, even the noise level. BUT the left side...

 

I should mention that I´ve cut out a paper ring to put it between the camera and the SB. The idea behind this was to make sure that there is no more space for the SB to move around. It actually worked, the SB didn´t move at all!

Unfortunately, the problem with the shifted in focus center (respectively out of focus left side) wasn´t solvable this way.

[Axel]

So what you suspect is that your Pocket's sensor is misaligned, don't you? Do you have another lens to compare? I had the Oly 12mm and the Nokton 25mm (sold them, which I rue now), and now I have the cheap 14-42mm kit lens. I know it's "through a glass darkly" compared to Sigma&SB, you can't focus comfortably (but you have a big DoF anyway). I recommend you borrow or buy such a cheap lens, before you actually change the Pocket. If the problem persists you need to send the camera to BM.

[sir_danish]

I don´t suspect the sensor to be misaligned, but the MFT mount on the camera.

 

I´ve tried two Speed Boosters and two Sigma 18-35. Both combinations produced the same image and the in focus center was shifted.

My other lens, the Lumix 14mm F2.5, is not the best comparison, because it is a rather soft lens and it only goes to 2.5 which produces a large DOF. At F2.5, the Sigma + SB look relatively fine too, because of the overall sharpness.

 

On my GH3, the Sigma looks fine, even wide open. I think I should get another very fast lens with Nikon mount and/or another BMPCC for a quick test. In the meantime I´m going to avoid wide shots at F1.8.

 

Axel, thanks again for your efforts!

[Axel]

I don´t suspect the sensor to be misaligned, but the MFT mount on the camera.

 

Somewhat academic distinction, but of course you are right.

 

My other lens, the Lumix 14mm F2.5, is not the best comparison, because it is a rather soft lens and it only goes to 2.5 which produces a large DOF. At F2.5, the Sigma + SB look relatively fine too, because of the overall sharpness.

 

You could do a test for me. Shoot brick walls asf. @ f3.5 - 5.6 with this lens in RAW, and tell me if you encounter moire (what I've found is that it need to be high contrast patterns, the old red bricks in my neighborhood won't show it. In my last images the balcony on the right edge shows vivid colors, but only with Sigma&SB so far, and only if I close the aperture). No need to publish the images. And while you're at it, if there is moire in raw, also shoot in ProRes. Thanks in advance.

[sir_danish]

I hope I can find a brick wall here in Graz...