Since 3X crop is sharper and very minimal aliasing and moire I went out and tried to see whats it like to shoot entirely in crop mode.
Used a Tokina 11-16 as a 33-48 "normal" lens and an 80-200mm nikkor as a very long 240-600mm.

Also, for this piece I removed the VAF filter to get optimum sharpness.

Quack 'n' Run - YouTube

...

Very nice!

The detail on grass and trees is very nice. The detail on the ripples on the lake seem to be the most difficult for YouTube's codec. I would guess that the original content looks much cleaner. What compression workflow are you using? We might also be seen limitations of the 80-200 lens. I'm not sure if the water is stealing compressed bits from the duck or if the duck is a bit soft in the original.

The scene at 0.42 is, um, interesting. The sun seems to have caused a magenta dot and larger area is blown out. This reminds me very much of some of the BM Pocket Cam footage with black dots and overflow of the blown out area. I've never seen this with the 5D2, but I realize I virtually never shoot toward the sun.

I wonder if the magenta artifact is due to the properties of the lens? Same with the overflowed area. Maybe this is flare blowing out the area, rather than a sensor thing. It would be interesting to shoot this at multiple exposures to find the balance between having enough DR for the sun without making the scene too dark and noisy. Of course, the magenta dot can be tracked and erased in post, and this would improve the shot by not giving a point to draw the eye.

Of course, the above critique is technical nitpicking. The footage is gorgeous compared to what we can achieve in h.264 and is nicely shot. Including the lake, grass, trees, and lens flare is a nice stress test. Your actor's skin and the lines on the boat look great.

BTW, have you shot much with a diffusion filter? I've got a Glimmerglass #1, which is quite nice for general use. It gives things a filmic look. (I've also got a #3, which is quite strong and needs to be saved for effects.) In a test that I saw recently the Digital Diffusion/FX filter is even nicer. While Glimmerglass has circular diffusion, the DigDif/FX smooths the image and gives more of a star effect on bright lights, so it has a more directional diffusion with less of a ghost look. Both can give a touch of that Hollywood sheen.

If I get a chance, I'll go out and shoot some similar outdoor footage with full and 3x crop on the 5D2 using the GG1 filter. I have the original Mosaic Engineering VAF, which I would use on FF only. The crop and lens length will definitely affect the diffusion.

Looks really nice. What version of ML are you now working with Ted ?

Chris,

there is now a Sep 4 build but for this piece I used aug 30 build. The sep 4 ver which comes after the 30 only has cosmetic changes.

Jon,

thanks for looking over the piece!

Very nice!
The detail on grass and trees is very nice. The detail on the ripples on the lake seem to be the most difficult for YouTube's codec. I would guess that the original content looks much cleaner. What compression workflow are you using? We might also be seen limitations of the 80-200 lens. I'm not sure if the water is stealing compressed bits from the duck or if the duck is a bit soft in the original.

I have gone back and tried a a.) larger file b.) different workflow and concluded that youtube loses so much detail. Case in point is the edging of my opening title. So yeah the ripple detail and duck detail are lost


The scene at 0.42 is, um, interesting. The sun seems to have caused a magenta dot and larger area is blown out. This reminds me very much of some of the BM Pocket Cam footage with black dots and overflow of the blown out area. I've never seen this with the 5D2, but I realize I virtually never shoot toward the sun.

I wonder if the magenta artifact is due to the properties of the lens? Same with the overflowed area. Maybe this is flare blowing out the area, rather than a sensor thing. It would be interesting to shoot this at multiple exposures to find the balance between having enough DR for the sun without making the scene too dark and noisy. Of course, the magenta dot can be tracked and erased in post, and this would improve the shot by not giving a point to draw the eye.

Yes, this is really a good stress test of everything. I have noticed that pink dot in a lot of my Sun light sources. I left it here unmasked for technical reference. At least its not a Black dot like the black magic magic pocket cam. It was there even in non crop mode raw shots. Sometimes it isn't there, look at the last shot. Next time I'll try to put out the zebras and see at what point it is triggered.

Also in post, I have noticed that in extreme lighting resolve can do a better job recovering highlights and balancing out the shadows than what After effects ACR can do.


BTW, have you shot much with a diffusion filter? I've got a Glimmerglass #1, which is quite nice for general use. It gives things a filmic look. (I've also got a #3, which is quite strong and needs to be saved for effects.) In a test that I saw recently the Digital Diffusion/FX filter is even nicer. While Glimmerglass has circular diffusion, the DigDif/FX smooths the image and gives more of a star effect on bright lights, so it has a more directional diffusion with less of a ghost look. Both can give a touch of that Hollywood sheen.

If I get a chance, I'll go out and shoot some similar outdoor footage with full and 3x crop on the 5D2 using the GG1 filter. I have the original Mosaic Engineering VAF, which I would use on FF only. The crop and lens length will definitely affect the diffusion.

I have read so many good things about these filters. Are they really good value for money? Would they be an essential part of ones kit?

Posting some frame grabs as I've realized Youtube really compresses the images and the improvement in detail with 3X crop can't be fully appreciated.

So readers will also get an idea. I've chosen scenes which would normally be problematic with aliasing and moire in 1X mode sans VAF, and artifacting with h264 compression.

Posted in Vimeo:

Quack n Run -shot entirely in crop mode with 5dmkII on Vimeo


probably better compression.

The jpegs do show more detail (see the trees) than even the HD Vimeo.

You did a nice job with the color, and I liked the lighting effects with the sun. But none of that has anything to do with RAW shooting. I do not see any advantage from your shooting RAW from this video - none. How could I, as there is no h264 comparison. I have seen stunning h264 videos from the same camera, so what?

Nobody in my experience notices how soft or sharp any video is, or how lacking in dynamic range, or how whatever until they see an alternative. And btw, I have shot RAW, and have come to the conclusion it is not worth it. You've shown it can be done in the field, but not why we should bother (not your fault). And if sharpness is what this is about, with less moire, the GH3 does better (again, if that is the only consideration).

Hi Mark,
This test is for the crop mode feature and how it handles situations that is otherwise difficult with the 1X, like aliasing for instance, and how it would look with certain lenses due to the crop.
The comparisons with h264 is already done and settled. Raw is a much more flexible medium and easy and fulfilling to grade.

The crop feature has been late to evolve due to framing and corrupt frames issues, its just been recently that it's usable to a degree so that's why I decided to do this and also a previous test I did showed that a VAF while very beneficial in removing aliasing and moire, it is not ideal in crop mode, so this time I removed it prior and shot the entire test without it.

And oh, the shots with the sun are so dark and contrasty ungraded, I don't think h264 would have achieved the latitude and detail I recovered from the image.
Here it is as shot:
DVXuser.com -- The online community for filmmaking (http://www.dvxuser.com/V6/attachment.php?attachmentid=76045&d=1378524068)

This was after the adjustments.
DVXuser.com -- The online community for filmmaking (http://www.dvxuser.com/V6/attachment.php?attachmentid=76044&d=1378523699)