Neutral Density filters for a 5D...
 

I'm looking to buy two 4x4 ND filter for my standard tray matte-box, and I wonder what would be the best price/performance. I'm also not sure if I should get two 0.3's, or one 0.3 and one 0.6, or perhaps one 0.3 and one 0.9, or...? I'm mostly gonna use them outside to get DOF when it's too bright. Any help appreciated.

-- peer

What ƒ stop do you want to get to?

If you're shooting in full sun and want to get to ƒ4 or wider you need much more ND than that.

Something in the 1.2 to 1.8 range would be what you're looking for.

Dave Smith

Peer:

Don't be a tightwad, spend the money, do it right. I have been extremely happy with it Singh-Ray Filters: Vari-ND Variable Neutral Density Filter Leave the 5D MKII at 1/60th ISO 100, dial in the DOF you want with your aperture, then control the look and exposure with the Vari ND. It's like having a second manual iris.

Well worth the bucks.

Dan

And sell my lovely super duper Century mattebox...?! Since it has two filter trays, why not use them instead of lashing out even more dough on a SinghRay?

-- peer

There are posts somewhere claiming that the lowest noise points are 160, 320, 640, 1250, and 2500 ISO. I haven't tested or confirmed it myself.

Anyway, with your variable ND, you could easily choose to hit 100, 160, or whatever you want, given enough light.

@Peer: Personally, I wouldn't choose two 0.3 NDs. If I only had one, I'd probably choose a 0.9 for the situation where I want to open the aperture by three stops in daylight. Few in the audience will notice a one stop change in aperture. On my lawnmower timelapse, http://vimeo.com/4697740 , I used a 0.9, closed the aperture to f/22, captured one second open and one second closed (the minimum settings on my controller), and I still blew out the highlights. I wish I had a 1.8 at the time. I could have exposed two stops lower and opened the aperture one stop to reduce diffusion.

So if I use, say, two 0.9 filters (3 stops each) on top of each other, will that result in 1.8 density... i.e., is it linear or is there some log factor involved?

-- peer

Yes, that's correct. In general a single 1.8 is better in that it will have less flare and color shift. Also, if you buy a 0.9 and 1.8, you can get three values: 0.9, 1.8 and 2.7 (stacked.) Many people buy 0.3, 0.6, and 0.9, allowing any value from 0.3 to 1.8 in one stop increments - that's eight values - if you're willing to stack up to three filters.

My mattebox has only two filter trays -- so do you think one 0.9 and one 1.8 would be sufficient in bright daylight to get me down to a 1.8 aperture without cranking the snot out of everything else?

Btw, do you have a good & affordable sucrose for 4x4 ND filters?

-- peer

Hey, Jon. This will probably sound like a stupid question. Am I better off to have the ISO set to 1250, rather than 1000? I'd get less noise with 1250? That doesn't make any sense at all to me. Maybe I'm reading what you wrote wrong, who knows. Thanks for your help, Jon. I like reading what you have to say on here. You're a rockstar!

I wanted to jump in and suggest that the difference between 1250 and 1000 is probably not perceptible. However, if your doing many takes in the same lighting condition it would be best to maintain the same ISO for all takes. That is where the ND filters hold additional value to allow more flexibility. I am still undecided if I will go with screw-on style or a full matte-box rig. I like the screw-on for more under-the-radar low profile shooting. Good Luck.

That's exactly what some are claiming. In fact, they would claim that 1250 is quieter than 800! Again, I haven't tested it, personally.

Here are some 3rd party photo examples: 5d Mark II noise tests, ISO 100 to 1000 - a set on Flickr

People are theorizing that this has to do with the balance between analog (native to the sensor) and digital gain. I think the explanation is that the analog gain jumps in full stop steps and is native to 160, 320 and so on. The other levels are synthesized digitally, or so goes the theory.

Also, turn off Highlight Tone Priority for lowest noise, though it can provide better highlight detail.

And I really enjoy reading and posting here. Chris runs a most excellent forum!

That's correct. ISO 160 is actually just ISO 200 with +1/3 EC. The camera darkens the image digitally to make it look like ISO 160, but 1/3 stop of highlights are clipped just as if you had added the +1/3 EC yourself. The reason it has slightly less noise is because of the increased exposure. The other down side *would* be increased posterization, except that the 5D2 already has far more bit depth than it can ever use, even with digital EC, so that is not a problem.

ISO 250 is actually just ISO 200 with -1/3 EC. The camera brightens the image digitally to make it look like ISO 250, which means there is no additional highlight headroom like you would have gotten if you did it yourself.

John Sheehy charted it all out here:

Headroom, footroom, and dynamic range of all the ISO settings on the 5D Mark II

Yes. The -1/3 ISO settings (160, 320, 640, 1250) are fine (they just have slightly less highlight headroom and slightly less noise). The +1/3 ISO settings (250, 500, etc.) are suboptimal because they have more noise and the same highlight headroom.

That's correct. Only Canon's 1D cameras and the original 5D1 had real analog gain for the third-stop ISO settings, but even then those gain settings had slightly higher read noise because they used a separate late-stage gain amplifier, not the same variable gain amp that does the full stop ISO settings. The 5D2 is like the rest of the non-1D cameras, including the 50D and 500D, in that it does the "tweener" ISO gains digitally. This can be known for a fact because the raw histogram is an *exact* match for what the pull/push would result in.

Agreed. Optimally, one would only use HTP for ISO 3200+, since there is no noise difference between ISO 1600 and 3200, it will provide the extra highlight headroom for free. But I find myself using it even when it increases noise needlessly, because I can't recreate the perfect nonlinear EC of the HTP in the Profile Editor (though I can easily do it in a raw converter).

Agreed.

very informative on how this camera behaves and why. Thank you.

Peer:

I guess it depends on what you are shooting. If you are shooting narrative, matte box and your glass filters all of the way.

I am shooting more documentary stuff, chasing subjects all over big city, subway, on the street. My huge matte box and $5,000.00 worth of Formatt and Tiffen glass would be a bit out of place, way too slow and clunky. The beauty of the VariND is that it is fast, low key and infinitely adjustable instantly.

If I shoot some of the shorts that I have in mind, I would use my matte box and follow focus. But for the doc I am on, the Vari ND is a Godsend, couldn't shoot without it.

Dan

I just found a much cheaper alternative to the Singh-Ray variable density filters on eBay. I thought that I would risk $100 & order one. A nice feature is that it is an oversize filter effectively already in a step-up ring to avoid vignetting on wide-angle lenses. e.g. the filter for a lens with a 77mm thread is 82mm in size. Also unlike the Singh-Ray filters they come in a much wider variety of sizes.

Fader ND Filter (from ND2 to ND400) 77mm - eBay (item 140325118069 end time Jul-04-09 20:20:00 PDT)

This appears to be the site of the manufacturer High Quality Professional Equipments - Fader ND filter (ND2~400)

This Hong Kong store that looks like it's the same eBay trader. High Quality Professional Equipments - Local Seller & Yahoo! ©ç½æ

That last site is mostly in Chinese but it looks like the prices are even cheaper than on eBay. Perhaps someone could help with some translation?

When I receive the filter I shall report back. If it's decent quality then at those prices I can afford to equip all my lenses. It's usually sunny here so any outdoor shooting requires ND filters if the lens isn't to be stopped right down.