Why is video from DSLR & Bridge cameras crap compared with a camcorder?

Video cameras are designed for maximum video performance. Still cameras create video with a software hack. Still cameras process video differently, too, probably smooshing the video to the highly compressed MPEG4 which must be unpacked for editing, usually. The compression algorithm is highly variable and usually suffers greatly when required to crunch weird images like low contrast shot in low light and highly detailed images like trees or moving filed of backlit grass.

My Nikon D7000 sees the whole 4k frame (about 12MP) when shooting video but, since only 1920x1080 pixels are used for video, not every horizontal is sampled individually, they get averaged, and the color range is reduced dramatically.

You can research image acuity on sites that compare image capturing, sampling,and video encoding and compression techniques.

In general, I'd agree with David's comments. It seems to me that for the most part, the excitement around DSLR is related to depth of field capabilities and and to the shooting opportunities that interchangeable lenses offer.

Having said that, I found the image quality of a camera like the 5D Mark iii – especially when shot with their inter-frame codec -to compare favorably to that from very good pro-sumer video cameras. Here is somecomparative videothat includes some that was shot RAW.

Russ

I would also point out that it is very likely the Panasonic SD800 has sharpening applied in camera during compression to storage media. This will definitely change how sharp the image is perceived to be.

You can apply a sharpening filter to DSLR footage to increase the perception of overall sharpness. Gentle use makes big changes in apparent overall sharpness of the source footage.

MtD

Another reason still image video often looks soft is because it is. If you are in Live View, focus is strictly manual and none of the DSLRs I've handled have focusing aids like my olde Nikons and Canons did. I'd LOVE to have a split image rangefinder or a microprism on the focusing screen!

The electronic shutter in a video camera is a slightly different animal than the rolling shutter processing of a full size sensor that is always on. IN older cameras, the little computer just could not keep up with the flood of data so some compromises had to be made. The result was less video data and a softer image. Newer camera have far more powerful processors.

Here are some references:

Encoder feed is 1720x974 (550D) and 1904x1072 (5D3).

Line skipping is one cause, upsampling is another, 422 to 420 is another, compression is another. Bitrate control only reduces compression artifacts.

Hello Erin,

There could be a variety of issues that could cause your video to look "soft". The first is focus -- DSLRs can have a very shallow depth of field, especially in lower-light situations. If you have your f-stop wide open (like f2.8), and you're zoomed in a bit, your depth of field can literally be just a few inches for clear footage. Which means that if your subject is moving around a bit, they will go in and out of focus as they drift towards and away from the camera.

What's your ISO setting? Higher ISO settings, which you would need to shoot as the available light decreases, will add noise and decrease the crispness of the video.

When you create your sequence, what settings are you choosing? If you were to pick a sequence resolution of (for example) 1280x720, but then export to 1920x1080, that would cause softness.

Are you shooting in fully manual mode, or are you letting the camera pick settings? If you're letting the camera pick the settings, you're not as likely to get the best results possible. If you're not that familiar with how the different settings affect your results, that's worth investing some time in learning. You might find this article I did helpful: http://www.churchproduction.com/story/main/how_to_shoot_better_footage_manu...

My best guess is it's a combination of the lighting levels that you're shooting under, the camera settings, and the focus/depth of field. The stock lenses that come with the cameras like the 60D (what I own for a DSLR) or the 70D are really hard to focus manually with great accuracy, which makes lower-light shooting for video a bit problematic. Try getting more light on your subject, and find a good balance between a higher F stop value to increase the depth of field, and higher ISO setting to get some for image brightness back up to where you need it to compensate for the higher F-stop value without adding too much noise into the picture.

Set up early before your next shoot if possible, pick what you think are the best settings, and shoot a short sample clip of you at the podium. Then, load it into your laptop and look at it full-res to see how it's coming out, and adjust your settings until you're getting the best you can get.

Hope that's helpful!

http://3d.coldstreams.com/?p=383

http://www.magiclantern.fm/forum/index.php?topic=2764.0

http://www.ninofilm.net/blog/2010/04/17/536/

http://www.ninofilm.net/blog/2010/04/26/full_hd_crop_zoom/

Interesting that both your still cameras record interlaced. I don't think I've ever worked with anything from a DSLR that wasn't progressive – typically, various models of Canon. Possibly why I've found the image quality to be comparable to decent video cameras ( so that I had no reservations combining a DSLR with the XF 200 in a multi cam shoot).

Russ

Interlaced? At the camera head? Hmm. My preference, and my experience and training, suggests shooting progressive at the camera and adding pulldown if interlaced is the delivery format.

To the interwebs!

When I used to hang out at the cow, the question came up often in various forms, usually in the mistaken idea that shooting 60i would produced 120 images and therefore allow some nice slow motion, even if it was at half resolution. Several of the major Cow wonks corrected the misperception: 60i is just 60 discreet frames recorded out as interlaced fields; it is not 120fps.

Shooting interlaced these days is an interesting idea but I don't see any benefits since most displays and performance venues will be projecting progressive.

Found this:

Normally for the same frame size, progressive will be have higher spacial resolution than interlace. To prevent aliasing the fields are created by using pairs of lines, so the upper field is created from lines 1+2, 3+4, 5+6 etc then the lower field uses 2+3, 4+5, 6+7 and so on. This line blending reduces the resolution slightly.

and this contrary point of view:

Progressive Video – I know that many people prefer to shoot progressive video, even on their video cameras, even if they are shooting for broadcast! If you have done a survey of posts on the internet you will see that there is a great debate going on about what is best, interlaced or progressive.

I have never seen the attraction to progressive video footage. Yeah if you freeze-frame a shot of interlaced video, it can have some nasty jagged edges, but how often does a video get paused, and who expects it to look good when it does? This has actually been one of my biggest objections to switching to DSLR. The DSLR cameras do not give you the option to shoot in interlaced. Most of my programs are broadcast or delivered on DVD. Both of these systems require that the video be interlaced. A general rule of thumb in video is that you should shoot, edit and deliver in the same format, whenever possible. Put another way, if you know that your video needs to be delivered on DVD or Broadcast TV, which needs to be interlaced, you should be shooting interlaced.

and this:

Almost all HDSLR cameras capture video using the progressive mode, which means that every frame is a complete picture. The alternative is interlaced video, which is actually twice the number of frames (60i or 50i) but each frame only contains half the image information, otherwise known as ‘upper’ and ‘lower’ fields. The two frames are combined on alternate lines to make up a complete frame, which produces smoother action video but can also leave a tearing effect. Interlaced video is more commonly used on camcorders than HDSLR cameras. Using progressive capture is, in effect, similar to using a very fast motor drive. The real advantage is that you can take out a single frame and create a still image from it, and you will have 30 or 25 frames to choose from for every second. Still images can be captured from interlaced movies but they need to be de-interlaced in order to remove any visible tearing caused by a slight mismatch between the upper and lower fields.

For further confusion:

http://dslrhd.com/2009/11/do-you-need-to-de-interlace-your-video/

http://www.dvinfo.net/forum/canon-cinema-eos-camera-systems/507532-25p-vs-50i-pr ogressive-vs-interlaced.html

http://www.wexphotographic.com/blog/beginners-guide-to-dslr-video

http://ediustips.com/index.php/observations/247-the-pros-and-cons-of-shooting-vi deo-on-a-still-camera

Ian R. Brown wrote:

I can't see how anyone could think 60i was 120fps as it is and always was just bog standard 30fps interlaced.

That's the bad thing about the Web. You can't go back and correct stuff, just extend the thread.

Telling someone that 60i is really "30fps but interlaced" is a tough sell. Add to the discussion that it's really 59.94 and 29.97 and you've got a fight on your hands.

Good luck on your film!

I shoot video with a Canon 5D mark III - the video is excellent and sharp, full of detail, and wonderful color.

The only problem I have is the awful compression YouTube applies to HD content which dumbed down color and led to a reduction in detail.

Just for fun, try a Vimeo upload to compare.

Russ