Compare FX and DX for video in low light

[danieljohansson]

<p>Hi!<br>

The difference between FX and DX is a topic that can puzzle any photographer, even me ;)<br>

In this case I want to figure out the exact difference in depth of field, noise etc and how it will effect the image for video in a low light situation? As an example I want to compare the Nikon D750 with the Nikon D7200 with a crop factor of 1,5.<br>

Of course you would think that the Nikon D7200 will give more noise in a low light circumstance but will it still be the case if you can adjust the aperture to get the same depth of field and lower the ISO? <br>

<br />Here is an example of two different gears and settings:<br>

Nikon D750<br />Focal length 24 mm<br />Shutter speed 1/50 s<br />Aperture f/8<br />ISO 3200<br>

<br />Nikon D7200<br />Focal length 16 mm <br />Shutter speed 1/50 s<br />Aperture f/5,6<br />ISO 1600</p>

<p>In my mind this would give the same result but is this correct or am I missing something...?<br>

Here are my simple calculations:<br>

24 mm = 16 mm for a DX sensor (24 / 1,5 = 16)<br />f/8 ≈ f/5,6 in depth of field for a DX sensor (more exactly it should be 8/1,5 = f/5,33<br />ISO 3200 ≈ ISO 1600 for a DX sensor (more exactly it should be 3200/1,5<em>^{<em>2 </em>}</em>= ISO 1422)</p>

<p>Will the footage look the same or will there be any difference in noise, brightness, depth of field, dynamic range etc? <br>

(sorry for any grammatical errors, I am not a native English speaker)<br>

Cheers!<br />//Daniel</p>

[ilkka_nissila]

In general if both cameras use similar technology, if you set aperture for equal depth of field, shutter speed the same,

angle of view the same, and ISO for equal luminance in the final image then you should get similar image level noise in

both formats.

 

However, video implementations may differ between the different sensors and thus it is best to compare the two cameras

in a real world application to evaluate the noise.

[Andrew Garrard]

<p>Your calculations are correct for still images. Sensor technology varies (for example, the D810, D750 and D5 all have substantially different dynamic range and noise management at different ISOs), but fundamentally having a sensor area 2.25 times as large (1.5x1.5) means FX is going to get roughly that much more light at the same (relative) aperture. Stop the aperture down to match the depth of field of the DX camera and the light advantage goes away. It's sometimes the case that a larger sensor can handle a bit more dynamic range through not running out of electrons in the pixels, and it's easier for a lens to resolve to the level of larger sensels (and generally easier to make a slower lens with FX coverage than a faster lens with DX coverage - and there aren't many DX lenses with the aperture to match an f/1.4 FX lens), so FX has some advantages - but the factors in your maths are usually dominant. The same applies to the DX crop mode of an FX sensor, obviously.<br />

<br />

Video is more complicated. Firstly, cameras use different crops of the image sensor in generating video. Then it's common for the processing to skip sensels (or at least lines) in generating video - this makes the resulting image look sharper (because it's aliased) but means only a fraction of the sensor is actually generating the image. Other cameras actually manage to downsample the whole sensor output properly, using all the sensel values - this is one reason Sony's A7S range is so good at video (in addition to being exceptional at low light stills). The D800 famously tended to skip a lot of sensor content compared with the 5D3, which made its low-light video look comparatively poor (I think - apologies if this is misinformation; it's certainly true of live view). If the camera isn't sampling all the pixels, the noise and dynamic range management will correspond to the amount of the sensor it's actually using - which could easily be "not a lot" for some cameras. With the same technology in use, for this reason, lower pixel counts tend to do a bit better - and I'd probably choose a D750 over a D810 for video if forced to either. The sharpness of the result can depend to some extent on the degree of downsampling, too - one reason I'd prefer Nikon to stick to something like the Sony 42MP sensor for their next camera rather than go for ridiculous resolution (the Sony sensor is slightly over 8K video resolution, and most of it gets used for downsampling to 4K or 2K video).<br />

<br />

Also bear in mind that only some Nikon cameras have "power aperture" (the ability to change the aperture during video). Doing this for a non-E lens is harder than for an E lens, since it requires independent control over the aperture lever when the shutter is open (present on the D3, D4 and D5 and on the D8x0 range; I actually don't remember about the D750, but the D700 can't do this). For video, that can be useful.<br />

<br />

While I've shot videos with my Nikon, the general rule for shooting <i>good</i> video is not to start with a Nikon camera, I'm afraid. While F-mount lenses get used in some places, companies like Panasonic, Sony, Canon, Red and BlackMagic are better options with more of a pedigree in video. I still have a BlackMagic pocket cinema camera on my wish list (partly because it can record 2K 13-bit raw without needing an external recorder). The Nikon bodies are fine for photojournalism or in an emergency, but if you want dynamic range or low light support Nikon have never really prioritised video.<br />

<br />

Says a man with very little video shooting experience, to be fair...</p>

[hans_janssen1]

<p>The aperture determines the amount of light that goes to the sensor and has nothing to do with fl and sensorsize. f8 + 3200 is the same as 5.6 + 1600. The dof has nothing to do with the exposure triangle, but is a side effect.</p>

[Andrew Garrard]

<p>(Relative) aperture determines the amount of light that hits the sensor per unit area - the same f-stop will allow 2.25x as much light to hit an FX sensor as a DX sensor (ignoring drop-off), just as cropping half of an image away halves the amount of light that went to that image. The exposure doesn't change - and ISO is a measure of sensitivity per unit area. But magnifying an image magnifies its noise, so comparing an ISO 100 FX image to an ISO 100 DX image blown up to the same size will make the DX image look noisier because the magnification is higher - and this magnification of noise effectively corresponds to the sensor running at the ISO needed to capture the same total amount of light. Everything normally cancels out (f-stop, as physical aperture divided by focal length, and ISO as a function of sensitivity per unit area), but you have to cancel <i>everything</i> - you can't say "an FX sensor is a stop better than a DX one" because it's capturing a different depth of field at the same f-stop, because it's using a longer lens, and if you compensate by changing the (relative) aperture you end up adjusting the ISO to compensate, so you're back where you started.<br />

<br />

So yes, f/8 at ISO 3200 is f/5.6 at ISO 1600 in terms of light capture - but both the aperture and the ISO affect the image.<br />

<br />

For a fixed camera position (so if we're not changing the perspective as well), the cone of confusion that determines depth of field in the scene is determined by the physical aperture (not the fraction of the focal length that the aperture is). So you get the same depth of field from a 150mm f/3 lens on an FX body as you do with a 100mm f/2 lens on DX - because they both have a 50mm entrance aperture. If you don't compensate for the crop by changing the focal length of the lens then things differ, but then you're not capturing an "equivalent" image anyway. Which I hope all agrees with Daniel's maths. :-)</p>

[ilkka_nissila]

<p>While Nikon's video quality at extremely high ISO settings may not be as good as other manufacturers, in normal conditions the video image quality seems very good. Remember that in typical uses of video, you use a slower shutter speed than would be used to capture motion-freezing stills. I typically use 1/200s or faster when photographing people indoors in available artificial light, but for video something like 1/50s should be fine, thus in practice you can get very good video image quality indoors in normal indoor lighting. However, holding the camera steady and focusing are challenges that may require the use of additional equipment to properly solve. I suppose my video requirements are relatively modest. We do use a GH4R with 25/1.4 for recordings in a very dimly lit room at work. However, there is something about Nikkors and the way the Nikon cameras form images (and video) that I personally find appealing.</p>

[bebu_lamar]

<p>Once reason many go for a DSLR for video instead of a regular video camera because the large image sensor. However, many of the DSLR crops the image and takes away the advantage.</p>

[Andrew Garrard]

<p>If the lighting is okay (and I agree it doesn't have to be <i>that</i> good), I have no objection to Nikon's results in video from my D810 - though I've not checked them for aliasing or similar. Obviously there's the bokeh from a large sensor (or at least the crop of it that's in use). But other options will record a higher bit rate if you want post-processing, including raw in some cases, and will likely do better moiré management. I'll happily use my D810 for video because I already have it (and have recorded several things using it - although most recently it's the time limit on video clips that I find annoying, and that's a legal nonsense), but for the right subject I'll just as likely use my RX100 or my V1. If I were buying a dedicated video camera, or a camera for which video was a major focus, Nikon wouldn't be near the top of the list - but I don't shoot enough video to object. Last time I researched it, though, I rapidly came to the conclusion that adding 10-bit raw recording over HDMI to a D810 using an external solution was going to be as costly and more inconvenient than just buying a Pocket Cinema Camera, even using a 4/3-to-F-mount adaptor (which it happens I already own), and I'd get better dynamic range from the PCC. Given how bad I am at managing light in still photography, leeway for digital correction in video seemed important - but if you know what you're doing or conditions are favourable then I'm likely worrying about nothing.</p>

[danieljohansson]

<p>Thanks guys for all the answers and opinions =)<br>

Since in my example with the Nikon D750 and Nikon D7200 are from about the same generation of sensors and I believe both cameras use the whole sensor for recording video my assumptions seems to be correct. I don't need to get into tiny quality differences for this case, just the overall theory =). And of course there are other features that matters for video like the ability to adjust aperture without turning off live view, power aperture, auto focus, flip screen, bit rate etc etc. The important summery for me is that there is no major advantage in using a full frame sensor for video if you don't need a short depth of field. <br>

If you are curios what's this is all about it is for shooting real estate videos. No high end productions but we are using 3-axis gimbals and some other gadgets. We also shoot stills off course and I am using a Nikon D600 for myself and it works fine for these productions. Just a bit annoying that I have to turn off live view to change aperture in video mode, but that's not a deal breaker since I am not filming every day. </p>

[ilkka_nissila]

<p>I found that for walking around to get a tour of an apartment with a video camera, 60p seems to look better than 24p since the latter shows a sequence of frames that you can see while moving (panning) around. 60p makes the footage look much nicer in this context, in my opinion. On the D7200 I believe you only get 60p in the 1.3x crop mode while the D750 offers that in full frame mode. Since you'll probably be using an ultrawide for this application, you may need to take into account the 1.3x crop in the DX camera (this crop is in addition to the DX crop) in your focal length choice to get the smoother looking footage at 60p.</p>

[danieljohansson]

<p>Good point about the 1,3x crop Ilkka!<br>

About the 60 fps, well even though the panning will look nice don't you think that it's better to stick with 30 fps because you will get shorter shutter speed and therefore lower ISO and less noise? </p>