Video vs Still - Intriguing Questions

[Lamar Lamb]

I got into digital video a couple of years ago never having done video before and having little experience with still image capture other than my Sony Mavica FD91. I bought two DSR-PD170's and was and still am amazed with the video they produce, especially in low light. Thoroughly enjoying my video work and wanting to explore the still side of things I purchased a Nikon D70 last June and have been devoting quite a bit of time to learning still photography as well. One thing that I saw immediately was that the digital SLR camera world holds onto film nomenclature. They use ISO for gain instead of gain.

Doing both of these has led me to ask a couple of questions. First off after reading discussions regarding lenses for HDV cameras and seeing the prices of some of these it makes me wonder why, when my $250.00 18 x 70mm lens does an outstanding job capturing a 3008 x 2000 pixel image, are the HDV lenses are soooo expensive? What's the difference?

Another thing that has me thinking as well is the low light gathering capability of the Sony PD170 image block / lens configuration vs the Nikon. Why is the Sony so good at low light. The Nikon is of course a single ccd at 23.7mm x 15.6mm at 6.24M pixels with a 67mm lens where as the Sony is 3 x 1/3" ccds. It looks as if the pixel size on the Sony ccd is bigger. Does that make a difference? I also would think the color reproduction on the Sony would be better than the Nikon due to the 3 ccd arrangement. Is the DSP better on the Sony? I know this is apples and oranges but it is intriguing.

[Rodney Compton]

Hi Lamar

I am an ancient stills photographer who has a liking for video and film, so I should be able to answer your questions. I shall try if I can.

ASA was invented by George Eastman. 100asa is equal to noon sunlight over the Eastman factory on mid summers day at a camera setting of
1/100th of a second at f10. At that time movie film was rated in exactly the same way, hence 1/25th second=(24fps) at f20'ish. Gain is an electronic term that the average real film tech would only understand in relation to electronic video cameras. As such Gain has an approximate equivalence to film sensitivity and speed. This meant higher ASA (or ISO) equals faster film, more grain and less image quality. So too with gain on a video camera. Boost the gain, degrade the image quality.

In respect of cost differentials on lenses, there might just be the matter of the massive numbers of stills lenses that are made compared with fewer made for video. There is also sales marketing within a specialised field: for example, I would have expected the floor to have fallen out of the XL2 market when the XLH1 was introduced - and it still may. In relative terms, though, if a stills camera is superseded, then the market in the older model drops dramatically. So, SLR lenses are consumer products, semi-professional video cameras are not quite that. Nevertheless, the complexity of a video zoom which covers a very large focal range compared with a stills lens, deserves to cost a lot, especially if it is built to a high standard. You may read about 'primes', well they are mainly good because they don't zoom; so too with still primes, they are designed for one focal length operation and for optimal image quality in a desired range.

In stills I have used the D100, the Canon 1DSMK11 and D20. Each camera has different characteristics, as do video cameras. I too am amazed how well modern colour control works in video cameras, especially the control of differing colour temperatures. Better I would say than stills cameras. The three CCD/Beam splitter arrangement on video cameras produces an RGB separated image on the CCD's, which as you rightly say should make for better colour.

The history of the 'beam splitter' goes back to Technicolor, where three separate cans of black and white negative film ran in synch through a movie camera registering the RGB images, these were later recombined to produce a colour film master. This form of excellent colour was standard in cinemas way after Tripack colour stock was introduced in the late 1930's.

Tripack is a process where all three colour layers exist in one film emulsion, and with development, this film type took over the whole market, both stills and movie. Modern digital stills cameras follow the same pathway laid down by Tripack film: they register the entire spectrum in one complex pixel array. The recombination of lifelike colour is achieved by electronic interpolation. Only the Foveon SIGMA D9/10 chip differs in registering colour through separate layers of RGB receptors.

There are always losses and gains when a machine does the thinking. In the old days with film, the cameraman had to determine all the cameras settings, this ability gave the cameraman control. It also made the cameraman aware of the limitations. These limitations drove development in film and camera design. Modern video cameras are products of many years of development, which is why they are so good, relatively. They are however, at the level of mini DV anyway, still compromising certain levels of quality for the sake of utility.

As a professional stills photographer, when I first picked up a video camera, I was interested primarily in image quality - particularly fidelity and sharpness. These criteria are important in certain fields, such as natural history, but quite unimportant in others, such as social documentary. So, just as in reportage, compared with still life photography, there is a fundamental divergence, so too with video.

The D70 is quite low on the list of quality cameras, but more than adequate to snap celebrities and other transient subjects, it is not however suitable for still life. Similarly the PD170 is ideal for all manner of socially oriented footage, but not for filming at long range or at the micro level. However, as with all issues of utility, some people will push the boundaries and make one thing do the work of another. The D70 could be used with a Micro Nikkor for close ups, but after having used the D100 extensively before using the Canon professional SLR, I know the D70 is inferior in more ways than just lacking several megapixels in output. Similarly, an interchangeable lens video camera 'should' be more flexible than one with a fixed lens system.

To conclude, I think the whole matter is about perception - i.e, what people expect to see. With serious photography people expect to see sharp, clear images. With TV and video, especially in the States - where the broadcast standard is only now in the process of redefining itself in terms of HD, people were not looking for ultimate levels of image quality. That is changing and HD will push the bar ever higher for the aspiring and the professional videographer alike.

Rodney Compton
naturefilms.tv

[Lamar Lamb]

I guess I’m coming into the still image arena backwards. I suspect most folks have still film experience first followed by digital and video whereas I learned on video and am now making the transition to stills. For me personally I’d like to have a digital still camera that was more like a semi pro video camera. I find the PD170 very easy to setup for full manual shots, much more so than the D70. You are right about the white balance. There are fewer presets on the PD170 but the manual white balance is much more accurate on the PD170. I know color is subjective but I think the PD170’s better. I favor the saturated “Sony” look anyway. I’m still working on finding a good setup in the D70 to get the color the way I like it without having to edit the pics in post.

I like the SLR and interchangeable lenses but am not sure why the mirror is still necessary. With the ability to view exactly what you get with good a crt or lcd viewfinder and having the ability to zoom in on a part of the image in the viewfinder image to focus (ala Z1U). Of course all this would add size weight and cost, I’m sure. But since were thinking outside the box…… I would think that this would be a better option than viewing directly through the lens. I have a hard time with manually focusing the D70 through the tiny viewfinder and use auto focus in most situations if there is enough light.

On a slightly different note: A funny thing happened last week. It had been a while since I had used my video cameras and I had been doing a lot of still work when I got a call to make a video promo for a cheer team. I went to video the team and while doing some handheld shots found it difficult to keep from turning the video camera to frame for portrait shots. I’d have to remind myself, “You can’t do that with video….”

[Rodney Compton]

Hi Lamar

I think Sony have just dispensed with the mirror on one of their SLR's.

I don't think you are exceptional in not having had stills experience. My experience tells me its 50/50.

There was a very good glamour photographer called David Hamilton who crossed over to film direction. Every shot in his film was a masterpiece - like a Degas canvas, but boy was the action slow. That sums up the difference between stills and movie thinking.


Rodney Compton

xyris.tv UK

[Rainer Hoffmann]

Originally Posted by Lamar Lamb

I know color is subjective but I think the PD170’s better. I favor the saturated “Sony” look anyway. I’m still working on finding a good setup in the D70 to get the color the way I like it without having to edit the pics in post.

Lamar,

unlike your video camera your D70 gives you access to 12bit uncompressed RAW data. Once you have discovered the beauty of RAW images and the possiblities of "editing the pics in post" you'll wonder why they don't give you access to the RAW data of your video camcorder (because of the sheer amount of data of course, but anway...).

If you shoot RAW images you just don't have to worry about white balance for example and even pictures underexposed by 1 f-stop (may be even more) can be restored without too much loss of quality. And if you like the color rendition of your PD170, well, it should be quite simple to find settings in your (then favourite) RAW processing software to simulate the PD170.

I suggest you give RAW a chance (wasn't that the title of a song??).

[Lamar Lamb]

I do the RAW thing now. I purchased Nikon Capture specifically for NEF format post processing along with some filter packages. I'm just trying to get to the point where post processing is unnecessary. I'd like to be able to go do a shoot, upload, delete the trash, then be ready to print.

I currently use Paint Shop Pro X for most editing. I know Photoshop is the de facto standard but I have been using PSP for 10 years and am familiar with it. In time if I find I can’t manipulate images to my satisfaction with PSP I will probably switch to PS.

I have been playing around with the custom enhancement settings on the D70 to get to that magic look I want but it has eluded me thus far. I need to spend more time tweaking.

[Rainer Hoffmann]

Originally Posted by Lamar Lamb

I'm just trying to get to the point where post processing is unnecessary. I'd like to be able to go do a shoot, upload, delete the trash, then be ready to print.

Ah, I'm afraid that will not work if you shoot RAW. It's the very idea of RAW that the data you get is not (or almost not) processed by the camera circuitry. Therefore, the settings of the camera won't have any influence on the RAW data but only on the JPEGs.

You can either have full control over your image data or accept the lower quality of JPEGs.

EDIT: If the D70 can store JPEGs and RAW at the same time, then you can always print those pics that you like the way they are right away and process the other RAW images later on.

[Andrew Khalil]

Hi Lamar,
I have also gone from video to stills but I don't think video cameras are better in low light, in fact, I think they lag behind still cameras a lot.

Video cameras have such tiny sensors that making faster lenses isn't very hard. Also, if you compare the grain in lower light images, DSLRs are much better.
I also think some of it is just the way we shoot with both - with video, using a lower shutter speed isn't a big issue because we don't really want to freeze motion. With stills, faster shutter speeds are necessary in order to prevent motion blur.

Optical viewfinders are necessary on DSLRs for a few reasons: First of all, LCDs on digicams are just too small and the resolution is too low to be able to focus properly. Combined with the larger sensor and therefore shallower depth of field, it doesn't make sense to put live LCDs on digital slrs. Also, keeping the sensor on all the time to create the preview causes it to heat up (giving you hot pixels) as well as making it really slow.

RAW data is something else video still doesn't have and it really wouldn't take up that much more space when you think about it - a RAW image from my Digital Rebel is about 6mb (depending on the scene). multiply that my 30 (if shooting 30P) and you have 180mb/second. But remember that even HD video resolution is much lower, about a 3rd of that, so it would be about 60mb/second and even lower than that for SD video. I think it would be possible, but also remember that video cameras have very little dynamic range compared to still cameras, another place they're lagging, again, mostly because of their tiny sensors.

I hope this helps

[Robert Mann Z.]

http://www.reel-stream.com/ gives you the ability to capture raw (unprocessed) feed from your dvx...

the idea works pretty much like raw from your still, a flat image that needs tending to in the post...

[Andrew Khalil]

yes, I've heard about the Reel-Stream, but I've honestly never heard from anyone using it so I don't really have much to say on it other than it looks really cool, but I wish cameras would just capture that much dynamic range without needing the mod.
I think another limit to video quality is the miniDV tape - I think cameras now have the ability to capture a lot more colour and detail than the MiniDV format allows and a new format is needed if we're to get more dynamic range. I think it's part of it, but I'm not an expert on all these little details so feel free to correct me if I'm off track.

[Lamar Lamb]

I didn't know that changing settings had no effect on the NEF files? About the only time I use the NEF's is for portrait work or shots I intend to modify from the start. I have not yet tried any setting changes with NEF's yet. Good thing I guess.... now that I know the answer to the question I would have had. :-)

I did know about the NEF + Jpeg but have not tried it yet. That is a good reason to use it though. I will try it out. It might help the work flow.

Originally Posted by Rainer Hoffmann

Ah, I'm afraid that will not work if you shoot RAW. It's the very idea of RAW that the data you get is not (or almost not) processed by the camera circuitry. Therefore, the settings of the camera won't have any influence on the RAW data but only on the JPEGs.

You can either have full control over your image data or accept the lower quality of JPEGs.

EDIT: If the D70 can store JPEGs and RAW at the same time, then you can always print those pics that you like the way they are right away and process the other RAW images later on.

[Bob Grant]

One thing I've noticed is a difference between moving and still images regardless of if they're film or video. Grain / noise is less noticeable in a moving image, I suspect our brains tend to either be distracted by the motion or just average it out. I suspect the latter is more the case.
If you take a locked off shot of a still subject and look at. Then take one frame from that and duplicate it to form a new video sequence the noise is far more noticable.
I should add a caveat to that, feeding noisy video through a system such as mpeg-2 that uses temporal compression. The compressor can run out of bandwidth trying to encode all those noisy pixels and what you end up with is noise frozen for the length of the GOP and then at the next "I" frame it jumps, very distracting to the eye. Film grain seems to go through the same process much better.
For this reason at times I've redone those locked off shots using a single frame, the noise is less noticeable as it doesn't jump every 8 frames or so.

[Rodney Compton]

Originally Posted by Andrew Khalil

yes, I've heard about the Reel-Stream, but I've honestly never heard from anyone using it so I don't really have much to say on it other than it looks really cool, but I wish cameras would just capture that much dynamic range without needing the mod.
I think another limit to video quality is the miniDV tape - I think cameras now have the ability to capture a lot more colour and detail than the MiniDV format allows and a new format is needed if we're to get more dynamic range. I think it's part of it, but I'm not an expert on all these little details so feel free to correct me if I'm off track.

Interesting isn't it, in 'stills' terms the Holy Grail has always been definition and contrast modulation through higher sensitivity, finer grain and more interlayers. However, movie film operators have always used negative to positive transfer, which is what, in stills terms, they dish out to punters through the drug store. No advertising photographer would bother with such material, preferring instead to use transparency film, which over the years has developed as many as twenty layers - each one tweaked to get every nuance possible out of a subject.

Needless to say, as a stills photographer using Kodachrome, when I first got involved with 16mm film, I was astounded by the rough old stock these movie guys used. Agfa was used a lot through the eighties by our company and although its colour was good, it's latitude was awful - maximum two stops if you were luck - before it burned out highlights. In reel-streem they are talking about pushing the boundaries of video footage latitude to 9.5 stops. I agree this is like using RAW files on a digital stills camera, but one of the points Lamar originally made, that none of you have addressed, is the superiority of using a beam splitter to sample the images through a video camera. Everyone takes it as a given, (me included), but a modern video beam splitter is a very sophisticated piece of kit when compared with a single layer pixel array in a digital stills camera and is undoubtedly superior at colour separation.

The downside in MiniDV, I have come to realise, is compression and until that is solved by faster processing and hard disk transfer, it seems we are stuck with what we pay for. However, compared with some of the 16mm stock I shot in the eighties, it's still amazing.

Rodney Compton

[Brendan Marnell]

Please allow me to dumb things down Rodney (I usually do) and ask you to clarify somethings for me that are probably obvious to you ...

I've been taking stills using 18-55mm and 75-300mm and 180mm lenses with Canon EOS500 and more recently Canon Digital Rebel for years. I'm videoing wildlife with Canon XM2 now, almost always handheld for run 'n gun. I'm thinking about the XLH1 and I'd love to know:
1. How you recognise that the XLH1 might supersede the XL2, apart from it being so much handier (especially for run 'n gun)? and
2. Would I be able to use my existing lenses on XLH1 including autofocus? or
3. If not, what would be their approximate equivalent lenses for XLH1?

Your value judgments I'm asking for, please?

[Michael Salzlechner]

a few differences

first off video quality is nothing compared to still quality so in general lenses as well as sensors do not need the same quality as a still camera does.

video lenses are built with long zoom ranges (10x, 15x, 20x) where good still photo lenses generally only have a maximum of 3x or 4x and some of the cheaper ones may get up to 10x (allthough in the still world we dont refer to them as such).

in addition video lenses are generally faster. a 20x F1.6-F3.5 zoom on a 35mm still camera would be almost impossible to create.

Now as far as low light ability still cameras are made to perform well at 100 or 200 ISO and then they do the best to get up to 1600 or 3200 ISO.

Video cameras use a sensor that is rated similar to ISO 400 or 800 at its base gain level because video cameras need to perform at lower light levels. That is why you need ND filters a lot more with video cameras as well.

Now if i compare some of my still lenses

28-70 2.8 L $1200
100-400 F4.5-5.6 L IS approx $1500
70-200 2.8 L IS approx $1600
300 2.8 L IS approx $4000
500 F 4 L IS almost $6000

Now obviously you can buy cheaper lenses but then you also end up with slower less quality glass as well.