XF300 sensor pattern visible?
 

On a brandnew Canon XF300 (PAL version), underexposed areas show a very thin vertical line noise-pattern. It's useless to post a screenshot, as the pattern subtly shines through only when the movie starts playing.

You can view it via my Vimeo account, but please note: the pattern isn't visible due to compression artifacts added after the upload. Please download the original file to judge.


Camera was set to +12db, Cinegamma 1, sharpness -3, 1080/25p. Same issue at +6db and even 0db, albeit less visible.

What does the camera perform like at a normal setting (0db)? I don't have high expectations for performance with that much that gain on any camera.

With that said, I don't see significant streaking, especially nothing like the streaking at high iso on the GH1. And I gotta say, the image is incredibly clean for 12db, in my opinion. Is this really 12db? Wow.

Lastly, I found that it's easier to see fixed patterns if the image is moving or something is moving within the image. I can't tell much from the static shot.

Please post more footage! :)

Hi Steev, thanks for your comment – yes, it is 12db. At 0db and properly exposed pictures the XF300 performs almost flawlessly in terms of noise pattern. I could upload a 0db and +6dp of the same shot if you like, but due to the twilight situation they are of course even more underexposed and the pattern isn't noticeable that much.

I'm not really complaining about the noise issue. I love noise, because noise gives the image an organic touch. But to see a fixed pattern within this noise makes me bother. I viewed the clip on a 30" Cinema Display and a MacBook (to ensure it's not the display's fault) in a dimly lit room, asked different people around my house and they all noticed what I saw.

Take a close look at the shadow area thrown by the sunshade on the MacBook enclosure. I will try to post more footage over the weekend.

Hey, Martin - I downloaded your footage and took a look. Then I did the same test with my camera at similar setttings (12db, cine 1, -3 sharpness). I do not get that vertical pattern at all, and I viewed on a 30" cinema display. You may want to have the camera looked at by Canon, that pattern may or may not be in spec. hard to say.

Brian, thanks a lot for checking! Does it mean that you noticed the vertical pattern in my footage? The Canon support here in Germany already took a look and proclaimed there is no pattern at all. Difficult position I'm in so far.

Maybe it's a phenomenon similar to the so called rainbow effect on DLP projectors: some people have „DLP-sensitive“ eyes and see it and others don't. Maybe it's just the PAL version. Reminds me of the day I discovered the PAL resolution bug in the HV30 PAL version. Hopefully more forum visitors will share their observations and maybe I will be heard the next time I knock at Canon's doors.

XF300 sample shot, now with moving image at +12db, 1080/50i. Note the vertical pattern on the chimney mantle and the white door's reflection on the floor. Please download the original file (which can also be viewed with vlc-player) and play at 100% size.

Well, yes, I can see it. Honestly, though, it's very hard to see - I don't think it would be visible on a television; cinema display, perhaps, but not on a tv. It may be that Canon considers that pattern to be in spec... I don't know. Could be a PAL issue. My NTSC version does not exhibit the same pattern. Since it seems to be the worst in underexposed dark gray areas, perhaps try to shoot an an underexposed gray card and bring it out as much as possible. Maybe shoot the same card at 0, 6 & 12 db with proper exposure & underexposure. Then, if you're local, take that to Canon and show it to them on your macbook screen, preferably directly from the CF card or file structure and see what they say.

Glad you could confirm my findings. Good to know that your NTSC version is not affected. I appreciate your proposal for shooting a gray card very much, and I will give it a try, but I've been rejected by Canon support once before. Probably I have to find another owner of a PAL version first and ask him or her to investigate. I tried to contact Nick Wilcox-Brown via Vimeo, but maybe he is on vacation. Again, Brian, thanks for checking.

I'm back from idyllic lakes, but as per email, I don't currently have and can't borrow an XF right now (I gather demand is very high and supplies are short).

Having viewed the clip, I personally think sensor is fine - a combination of underexposure and gain gives an effect more like +18. If I can get a camera, I'm really happy to try and match the test to compare (and it would be good to meet up with Alister too!)

Nick.

Better make it quick as I'm off to India and Singapore for some 3D workshops, then its IBC and so on.

Yes, I've got anther holiday and then work in the ME before IBC. Hopefully shooting with a friend's 305 next week, so I may get to try the pattern with that.

Nick and Alister, thanks for your effort! So far I couldn't find any NTSC footage on the internet which shows the pattern, maybe because either the XF300 NTSC version is not affected (see Brian's reply) or the footage was shot without gain (+6db or +12db) engaged. My camera exhibits the vertical pattern in slightly underexposed picture areas when shooting at twilight. I'll be on vacation for the next couple of days and hope to hear from you soon.