Working with 24p & deep DOF -- some postive ideas. - Page 2 at DVinfo.net
DV Info Net

Go Back   DV Info Net > Sony XAVC / XDCAM / NXCAM / AVCHD / HDV / DV Camera Systems > Sony HDV and DV Camera Systems > Sony HVR-V1 / HDR-FX7
Register FAQ Today's Posts Buyer's Guides

Sony HVR-V1 / HDR-FX7
Pro and consumer versions of this Sony 3-CMOS HDV camcorder.

Reply
 
Thread Tools Search this Thread
Old October 22nd, 2006, 05:59 PM   #16
HDV Cinema
 
Join Date: Feb 2003
Location: Las Vegas
Posts: 4,007
Quote:
Originally Posted by Mikko Lopponen
Yes, resolution will go down 50% in moving scenes but that resolution isn't in the interlaced file either. How can you say that bob deinterlacing loses resolutions when it simply displays one field after the other? There's NOTHING lost compared to the original interlaced material.
You are simply incorrect. The fact that interlacing is used does NOT mean a frame does not have approximately 700-TVL (V1). It makes no difference if the object is moving or is static.

When bob is used one field is discarded. Gone! And, with it goes half the information in a frame. Resolution is cut in HALF.

The progressive display, if 1080-lines, is now needs 540-lines to replace the 540-lines discarded. It gets them by copying the 540-lines that were not discarded and placing them in the frame. Resolution is not gained.
__________________
Switcher's Quick Guide to the Avid Media Composer >>> http://home.mindspring.com/~d-v-c
Steve Mullen is offline   Reply With Quote
Old October 24th, 2006, 01:14 PM   #17
Major Player
 
Join Date: Mar 2006
Location: Helsinki, Finland
Posts: 393
Quote:
Originally Posted by Steve Mullen
You are simply incorrect. The fact that interlacing is used does NOT mean a frame does not have approximately 700-TVL (V1). It makes no difference if the object is moving or is static.

When bob is used one field is discarded. Gone! And, with it goes half the information in a frame. Resolution is cut in HALF.
Google for how bob-deinterlacing works. I've been doing slow motions using bob-deinterlacing a lot and it really works wonders. There are dozens of deinterlacing methods available on the pc with different softwares and bob is simply the best for video.

Bob doesn't discard one field because the other field is displayed immediately afterward. It shows the fields one by one. NO FIELD IS DISCARDED in bob-deinterlacing.

Quote:
It makes no difference if the object is moving or is static.
You're actually complaining about losing 50% resolution in moving objects without comprehending that it's just a fact of interlacing. How are you going to combat against that? Your 10 000 dollar deinterlacers are just like magic bullets deinterlacer or alparysofts in that they make stuff up. They calculate the movements of pixels and make up stuff that isn't in the interlaced material to create something that looks progressive without interlace twitters. Bob doesn't, it shows you exactly how the material would look like if it would be played in an interlaced monitor. Except 30i is converted to 60p.

AVISynth Bob deinterlacer is one the best deinterlacers out there in the pc world. Try it out and compare it to...well anything really.

Bob is also used in a lot of pc dvd software like windvd and powerdvd.

There are a couple of sites that regularly complain that bob-deinterlacing loses 50% resolution. That is wrong. They don't take into account the fact that in bob every field is shown. They're not discarded even though one frame will be half. The other frame will be the other half. And as the images are shown at double the speed (30i -> 60p) that means nothing is lost compared to a normal interlaced monitor that also shows line after line after line.
Mikko Lopponen is offline   Reply With Quote
Old October 25th, 2006, 01:25 AM   #18
HDV Cinema
 
Join Date: Feb 2003
Location: Las Vegas
Posts: 4,007
Quote:
Originally Posted by Mikko Lopponen
NO FIELD IS DISCARDED in bob-deinterlacing.
If you want to claim your PC experiments prove everyone else is wrong about bob deinterlacing -- there's no way to stop you.

But, you might want to consider why published experts are wrong and you are right. It doesn't matter what you do on your PC, you can't change the definition of deinterlacing.

bob discards one field of every two. The second field is never shown. That is where your error lies.

It is weave that uses both fields.
__________________
Switcher's Quick Guide to the Avid Media Composer >>> http://home.mindspring.com/~d-v-c
Steve Mullen is offline   Reply With Quote
Old October 25th, 2006, 01:57 AM   #19
Trustee
 
Join Date: Mar 2004
Location: Milwaukee, WI
Posts: 1,719
Steve,

A weave shows both fields at the same time or in other words it doesn't do anything at all. It just weaves field 1 and field 2 together into a frame which can show interlacing on moving objects.

A bob works just like it sounds. It alternates fields into frames which can cause small details to bob up and down due to the alternating lines each frame. Bob is the process of turning 60i into 60p. This is listed on many websites and in a lot of software programs. VLC player uses the bob mode to playback interlaced material as 60p.

Deinterlace is just that. Just throwing out a field or removing the second field. Better known as discard.

A blend will bob the frames and then blend every two frames together to form a 30p sequence. Kind of like the motion filter on the JVC HD-100.

There are only two ways to get better than a bob. One is to only bob the sections that move and duplicate the non moving pixels. Kind of a blend of bob and weave. The other method is motion adaptive which is the huge beast to try and process.

Just check out how VLC player does it.
Thomas Smet is offline   Reply With Quote
Old October 25th, 2006, 05:53 AM   #20
HDV Cinema
 
Join Date: Feb 2003
Location: Las Vegas
Posts: 4,007
Quote:
Originally Posted by Thomas Smet
Steve,

A bob works just like it sounds. It alternates fields into frames which can cause small details to bob up and down due to the alternating lines each frame. Bob is the process of turning 60i into 60p.

Deinterlace is just that. Just throwing out a field or removing the second field. Better known as discard.
By your definition, bobing isn't de-interlacing but a frame-rate doubler. However, other definitions have bob discarding every other field which is "bob de-interlacing." If the frame is larger than a field, then the remaining field must be scaled.

http://www.100fps.com/

A paraphrased quote to be legal:

"Line doubling takes the lines of each interlaced frame (ONLY even OR odd lines) and doubles the number of lines, filling the entire frame, resulting in video with effectively half the vertical resolution. This technique is also called bob deinterlacing."


A paraphrased quote to be legal:

"Deinterlacing assembles progressive frames and shows them twice to use up the same amount of time as two fields. The need for 60 flashes on the screen each second stems from the Flicker Fusion Frequency."

Most HDTVs deinterlace. They discard, as you correctly say, every other field. The discarded field is gone and with it 50% of the frame's resolution. That is the way to eliminate interlace artfacts. The question then is by what process the missing lines are created. There are many ways and they are not hard to do with a DSP. 2D FIR is fine. Line-interpolation or line-doubling (scaling) is the simplest.

When you read an HDTV person use the term bob, they mean "bob deinterlacing."

Of course, HDTV's don't need to deinterlace. As you point-out, both fields can be combined. This is weave. It can be used only for static images, or even for motion with very heavy filtering to remove combing.
__________________
Switcher's Quick Guide to the Avid Media Composer >>> http://home.mindspring.com/~d-v-c
Steve Mullen is offline   Reply With Quote
Old October 25th, 2006, 07:25 AM   #21
Trustee
 
Join Date: Nov 2005
Location: Sauk Rapids, MN, USA
Posts: 1,675
Quote:
Originally Posted by Steve Mullen
...
- there really wasn't a quote that addressed the "keep the deep DoF" piece of the first post here, but that's what I'm addressing.

One of the problems with video is that you're working with a more limited resolution medium. If you shoot Deep DOF (which is a perfectly legitimate artistic choice), the backgrounds get mosaiced into the little squares of the ccd...if you have a background object that only spans 2 pixels, but is high contrast, it starts to do funny things if it moves. These artifacts of the non-random, small pixel count are not visually appealing and are the primary cause (the other being the limited exposure range) of the "video" look.

I personally am annoyed by the "film look" thing as a whole...but that isn't to say we need to discount the techniques employed to that end. Arguments will be made that there have been tons of deep DOF things shot and shot well...but the material that is constantly referred to was shot on a much higher resolution format...even s8mm footage has a resolution more comparable to that of HD than DV. And s8mm has a "technically" shallower DoF than most HD/DV cameras. A 1/3" CCD is really only 7mm. The larger the capture plane, the shallower the DoF at equivalent iris size.

I see selective focus as a tool, I see deep DoF as part of that same tool kit, but on lower resolution, grid based capture media, it's too easy to end up with the audience looking at the artifacts caused by the format rather than concentrating on the story being presented.

What makes a good picture is shooting within the limitations of the format. So light well, expose correctly (DV is less forgiving of bad exposure than film) and even softening the background slightly can make a dramatic impact on the apparent quality of your footage.

As I've stated in the past, the "Film Look" happens in front of the camera, not in it. Like a paint brush, the mechanics and physics of the camera is painfully simple stuff...the art happens when that tool is applied in a creative way to produce an image that tells a story effectively.
__________________
Web Youtube Facebook
Cole McDonald is offline   Reply With Quote
Old October 25th, 2006, 07:46 AM   #22
Trustee
 
Join Date: Mar 2004
Location: Milwaukee, WI
Posts: 1,719
Quote:
Originally Posted by Steve Mullen
By your definition, bobing isn't de-interlacing but a frame-rate doubler. However, other definitions have bob discarding every other field which is "bob de-interlacing." If the frame is larger than a field, then the remaining field must be scaled.

http://www.100fps.com/

A paraphrased quote to be legal:

"Line doubling takes the lines of each interlaced frame (ONLY even OR odd lines) and doubles the number of lines, filling the entire frame, resulting in video with effectively half the vertical resolution. This technique is also called bob deinterlacing."


A paraphrased quote to be legal:

"Deinterlacing assembles progressive frames and shows them twice to use up the same amount of time as two fields. The need for 60 flashes on the screen each second stems from the Flicker Fusion Frequency."

Most HDTVs deinterlace. They discard, as you correctly say, every other field. The discarded field is gone and with it 50% of the frame's resolution. That is the way to eliminate interlace artfacts. The question then is by what process the missing lines are created. There are many ways and they are not hard to do with a DSP. 2D FIR is fine. Line-interpolation or line-doubling (scaling) is the simplest.

When you read an HDTV person use the term bob, they mean "bob deinterlacing."

Of course, HDTV's don't need to deinterlace. As you point-out, both fields can be combined. This is weave. It can be used only for static images, or even for motion with very heavy filtering to remove combing.
Dude in that same site you listed did you happen to notice that under bob it mentions it has very smooth movement and a need for a system to be able to playback 50fps? That is where I took the definition of what bob is in my other post.

If all of these HDTV's are doing just a deinterlace and throwing out a field then basically all they ever show with interlaced material is 30p if what you are saying is true. That is not true at all. Since a lot of people complain about 30p from the JVC-HD100 I'm sure it would have been a huge issue if a HDTV could only show 30p. It wouldn't look very real to a lot of people. Every HDTV I have ever looked at when I was looking to buy one looked like it was showing smooth 60hz motion even on 1080i channels. I have never in my entire life ever saw a HDTV that only showed 60i material as deinterlaced 30p.

Couldn't the test be thinking of a normal 60i to 60p bob as a failure since it doesn't try to enhance the missing detail? In fact I have a Samsung HDTV on that list that has failed the deinterlace test but it is still converting 60i to 60p. I know this for a fact because I have edited a HDV tape with different conversions where one version was 60i and the other was 30p. The 60i clearly flowed better then the 30p version which means my failure of a TV was still showing all 60 fields.
Thomas Smet is offline   Reply With Quote
Old October 25th, 2006, 09:40 AM   #23
Trustee
 
Join Date: Jun 2004
Location: Denver, Colorado
Posts: 1,891
Quote:
Originally Posted by Thomas Smet
If all of these HDTV's are doing just a deinterlace and throwing out a field then basically all they ever show with interlaced material is 30p if what you are saying is true. That is not true at all. Since a lot of people complain about 30p from the JVC-HD100 I'm sure it would have been a huge issue if a HDTV could only show 30p. It wouldn't look very real to a lot of people. Every HDTV I have ever looked at when I was looking to buy one looked like it was showing smooth 60hz motion even on 1080i channels. I have never in my entire life ever saw a HDTV that only showed 60i material as deinterlaced 30p.....

..... In fact I have a Samsung HDTV on that list that has failed the deinterlace test but it is still converting 60i to 60p. I know this for a fact because I have edited a HDV tape with different conversions where one version was 60i and the other was 30p. The 60i clearly flowed better then the 30p version which means my failure of a TV was still showing all 60 fields.
I completely agree! Something is wrong in the interpretation because it is manifestly evident that every other field is not being discarded, else it would end up looking like low rez 30p frames versus smooth movements recorded every 1/60 second as fields.
Tom Roper is offline   Reply With Quote
Old October 25th, 2006, 07:36 PM   #24
HDV Cinema
 
Join Date: Feb 2003
Location: Las Vegas
Posts: 4,007
Quote:
Originally Posted by Thomas Smet
Couldn't the test be thinking of a normal 60i to 60p bob as a failure since it doesn't try to enhance the missing detail? In fact I have a Samsung HDTV on that list that has failed the deinterlace test but it is still converting 60i to 60p.
There are three facts:

1) 60i gets converted to 60p

2) the 60p motion is smooth

3) Less than half HDTVs are able to offer max. vertical resolution while more than half of the HDTVs limit resolution to about 50% of this value.

The reason for this situation, is the fact that interlace video must be deinterlaced. Depending on HOW it is done, there may be a reduction in V rez. The HDTV's that show the reduction use a method called "bob deinterlacing."

=========

UPDATE:

When bob de-interlacing is used, EACH field is input and placed into a frame. (You were correct -- one field is not discarded.) The frames are displayed at 60fps. This is the 60i to 60p conversion.

1) If the frame has 540-lines, as was the case with EDTV's that had 480-lines, then the field is used as is. This is common in software.

2) If the frame has between 540-lines and 1080-lines, then up to 540-lines have no information. These cannot be left black otherwise the brightness would drop by up to 50%. So up to 540-lines must be "created."

With bob deinterlacing, scaling is used to obtain the missing 540-lines. If 540-lines are needed, then this scaling is typically called "line-doubling."

bob deinterlacing uses "interpolatation" so the new lines are created as an AVERAGE of the lines above and below them. Interpolatation may reduce resolution by up to 50%. Why?

The test pattern has areas where very fine (1 pixel) black and white lines alternate. The pattern also has a moving object that will trigger the HDTV to use whatever type of deinterlacing used for video with motion.

The MT story states (paraphrased): "TVs that fail the test, show strobing black and white "boxes" because they double the first field that consists of just white lines, and then double the next field that has only black lines. Each interpolation of just odd or even lines wrongly assumes that, if odd lines are black, then the missing lines must be black. When a white field appears, it makes the same wrong interpolation, causing the box to appear to be all white. This causes a 50% loss of resolution in EVERY frame. However, when bob-deinterlacing is not used, the boxes show the alternating B&W lines as they originally appear."

Therefore, bob deinterlacing inherently causes a loss of effective vertical resolution of each field's original resolution.

Thus, we are both correct. Fields are not discarded, but HDTVs that use bob-deinterlacing reduce resolution up to 50%.

Which means that there is a 50:50 chance that when your interlaced production is actually viewed -- the resolution of the entire screen will be reduced by up to 50% whenever there is motion ANYWHERE on the screen.

Although 50% loss are the test results, with real video the loss should be only about 25%. For 1080i video that would mean the scaler output would average about 750-TVL and that was fine when displays were only 720- or 768-lines. It's really showing up as displays get 1080-lines. So that may be why so many folks don't notice it.

1) The loss is NOT from interlaced video as claimed.

2) It is from bob-deinterlacing despite what was claimed.

3) The fact that three brands had all their HDTVs pass the test means that when bob-deinterlacing is NOT used, resolution is significantly increased.
__________________
Switcher's Quick Guide to the Avid Media Composer >>> http://home.mindspring.com/~d-v-c

Last edited by Steve Mullen; October 26th, 2006 at 06:33 PM.
Steve Mullen is offline   Reply
Reply

DV Info Net refers all where-to-buy and where-to-rent questions exclusively to these trusted full line dealers and rental houses...

B&H Photo Video
(866) 521-7381
New York, NY USA

Scan Computers Int. Ltd.
+44 0871-472-4747
Bolton, Lancashire UK


DV Info Net also encourages you to support local businesses and buy from an authorized dealer in your neighborhood.
  You are here: DV Info Net > Sony XAVC / XDCAM / NXCAM / AVCHD / HDV / DV Camera Systems > Sony HDV and DV Camera Systems > Sony HVR-V1 / HDR-FX7


 



All times are GMT -6. The time now is 02:37 AM.


DV Info Net -- Real Names, Real People, Real Info!
1998-2024 The Digital Video Information Network