Canon 7D Mark I

[Audionut]

dfort continued development work on 7D over here.

[mblackshore]

I got that the 7D records at almost 1080P

1728 x 972 = 1679616 Pixels in total 16:9

I would like it to run 1920 x 874 which is  a aspect ratio of  2.2:1 a total pixels: 1678080 (a bit lower than the settings above at 16:9)
Is this possible without  entering crop mode?

[Walter Schulz]

Easy! Just replace 7D's sensor (+processing units +firmware) with one having at least 1920x3=5760 pixels horizontal.

[mblackshore]

Easy! Just replace 7D's sensor (+processing units +firmware) with one having at least 1920x3=5760 pixels horizontal.

AHA, it's a hardware limitation! Got it!

[Wlad81]

I would like it to run 1920 x 874 which is  a aspect ratio of  2.2:1 a total pixels: 1678080 (a bit lower than the settings above at 16:9)
Is this possible without  entering crop mode?

You can easily make Full HD out of what you have now by upscaling the video using MLV App. No one will notice the difference between the true 1080p and the upscaled one.

[reddeercity]

.... make Full HD out of what you have now by upscaling .... using MLV App. No one will notice the difference between the true 1080p and the upscaled one.

Yes you can , it looks really badddd !
Line skipping sensor don't up scale very well at all , it create more artifact then the original file .
Best to use Crop_Mode and record @ 2k(or 1080p) it's non line skipping (even line used)
You just need to use a 15mm-24mm lens to get the same framing as a 50mm for example .

[mblackshore]

Yes you can , it looks really badddd !
Line skipping sensor don't up scale very well at all , it create more artifact then the original file .
Best to use Crop_Mode and record @ 2k(or 1080p) it's non line skipping (even line used)
You just need to use a 15mm-24mm lens to get the same framing as a 50mm for example .

Your suggiesten about the lens was brilliant idea acually.
If I would upscale I would do it in post with Resolves SuperScale using neural network.

BTW does that mean that the built in video recording function with H.264 us upscaled output since sensor does not cover the size? Or is it that ML only can use X amount of the sensor?


How do you deal with noise artifacts when recording with raw? I seem to get allot more artifacts when recording with raw than not. There are artifacts that looks similar to chromatic abberation but not really. I will try to post some tests I have done tomorrow for you guys to see.

[Wlad81]

Best to use Crop_Mode and record @ 2k(or 1080p) it's non line skipping (even line used)
You just need to use a 15mm-24mm lens to get the same framing as a 50mm for example .

One not only just needs a wide angle lens, but also a lot of light, and this is not always possible.

[Walter Schulz]

Are you trying to say crop and non-crop mode behave different in regard of lens aperture or sensor photosite sensitivity?

[Wlad81]

Yes, that's it. The lesser crop is and the more lens aperture is, the lesser noise we would have. Considering the fact that the sensors that were used in those Canon cameras are quite old and aren't backlit sensors - the BSI increases the sensor's sensivity.

[Walter Schulz]

Sorry, only valid to lens aperture but crop has no influence on exposure. Easy task: Print a pic. Print a crop of a pic. Same exposure.
Proof or I'm calling BS.

A lens' T-number doesn't change if a smaller or bigger area is exposed to light.

[Wlad81]

The crop of a finished pic is not a whole picture taken with a smaller part of a sensor. The more is the crop, the more light and/or the more sensitive sensor we need to achieve the same s/n as with no crop.

[Walter Schulz]

BS.

[Wlad81]

I've got a lot of noise in the shadows when shot with crop 1080p mode.

[Walter Schulz]

I get a lot of noise in dark areas with all kind of sensors, lenses, apertures, focal length, exposure durations, crop factors.
Proof your point or I call BS.
At the moment you carry a load of BS.

First call: You are trying to tell there is a difference if a pic is cropped in post or if the area is cropped during aquisition. Total BS.

EDIT: Maybe it is beyond your comprehension how a sensor works: Imaging yourself as a photosite. And you are looking into the sky. And around you are several millions other people. Now tell all the people more than 500 meters away from you to close their eyes. Does your vision gets darker? Nope. Because it doesn't affect your eyes if other people far away from you have their eyes open or not.
If you believe otherwise: Give proof or I call it BS again.

[theBilalFakhouri]

I've got a lot of noise in the shadows when shot with crop 1080p mode.

In none-crop 1080p mode more pixels are being used (more photons are being captured), there is a horizontal pixel binning in this mode, resulting in less noise.

[Wlad81]

theBilalFakhouri, this is what I'm trying to tell.

[reddeercity]

In none-crop 1080p mode more pixels are being used (more photons are being captured), there is a horizontal pixel binning in this mode, resulting in less noise.

If you are talking about Line skipping sensor's I disagree , the that's what cause the noise , the pixel binning
https://thedailynathan.com/demosaic/algorithms.php?algorithm=pixelbinning&image=raw.png

Pixel binning is the most straightforward way to combine the color information across multiple photodetectors.
We simply assume a larger size for each pixel (each pixel is now an aggregate of information from multiple photodetectors rather than just one) and combine the color information

So yes you have more pixel , but the color/RGGB data  is not the same as a single pixel RGGB , It's being combine and that's where the noise being produce.
I done many noise level test and the cleanest is crop_mode on D4(5d2 & 50D) .
but you don't have to take my word for it ,  test it for your self .

On the 5d3 there's not line skipping so this doesn't apply , just cams like 7d , 5d2 , 50d etc. ....