|
|||||||||
|
Thread Tools | Search this Thread |
April 10th, 2004, 07:05 AM | #1 |
Major Player
Join Date: Dec 2003
Location: warsaw, poland
Posts: 440
|
pixels vs lines
ok, i cannot explain this to myself. so, i need help.
can someone explain me definitelly the differences between pixels and tv lines. let start with SD television - in pal countries you are always hearing about 625 lines (525 in NTSC), but the picture size is 720x576 PIXELS. (or is it 768x576?) so what is what - are the lines (625) just the full amount of lines we are using to TRANSMIT signal, and finally we have 576 lines/pixels for the picture only? or what? sorry if this was dicused earlier - if yes - please send me any links. thank you. filip |
April 10th, 2004, 11:44 AM | #2 |
Wrangler
Join Date: May 2002
Location: Vallejo, California
Posts: 4,049
|
For NTSC video, the 525 lines are not all visible as a fixed number are used for Vertical retrace (of the electron beam) during which other uses like a synch signal, closed captioning, etc. are generated. Only 480 lines are meant to be visible. (well, a few more actually but in NTSC DV, it is 480) PAL has a similar ratio of total to visible scan lines.
In this case you can directly equate lines and pixels. Horizontally we talk about 720 'lines' of resolution. In the NTSC DV world, we use that to mean 720 pixels because it works for us. That's all you really have to know to operate quite well. ------------------------------------- But there is no way you can see the detail that 720 by 480 theoretically can represent. What you might want to know is that in conventional resolution terms, the 720 by 480 pixel field of television isn't very useful as a description of how the human eye can see a displayed image. First of all, an all-black or white or gray or red or .... 720 by 480 image is just a plain solid-color field. If you want to understand the detail capability of a display, you need to understand the line-pair resolution of that display. A line-pair is a black line immediately adjacent to a white line. Full-black and full-white. Great! We can have 240 by 360 line-pairs, right? Usually not. If you put up an image that is a set of 240 horizontal line-pairs, what you will see is sort of a gray field that doesn't look quite uniform. That's because the display doesn't create razor-sharp edges to the displayed pixels. The edges appear gray and only the center (may) be solid black. Same thing for the white space (only in reverse). Now if you start reducing the number of line pairs, at some point you will get the display to deliver an image that 'clearly' is composed of alternating black and white lines. When you read the specs on a display for which they publish the line-pair resolution capability, they usually define a line-pair resolution at which the black and white spaces reach 70% of their maximum values. You have to read their spec. to know. Note that this ignores the issues of display addressability. We can address 720 by 480 pixels in our DV systems. But the real world isn't set up with fixed-location pixels. They can be anywhere in the visual field. So when we capture or place an image into our digital visual field, it can only be placed where there is a pixel. And they are rigidly organized in rows and columns. So image details usually get moved around a little bit.
__________________
Mike Rehmus Hey, I can see the carrot at the end of the tunnel! |
April 10th, 2004, 03:21 PM | #3 |
Major Player
Join Date: Dec 2003
Location: warsaw, poland
Posts: 440
|
mike, thank you for your answer, but i'm still confused.
so, to clear my mind - here are the questions: as i understand - if we are shooting a picture which consist in complicated patern, like mozaic or so, and by chance this mosaic is 720x576 tiles (say, in pal :))). now, we are making compositionon our PAL system in wievfinder, and the whole masterpiece is in the frame. good. but! if we are shoting this mosaic masterpiece, we will not see the whole tiles correctly, because the camera is probably not adjusted in perfect "locked" - 1:1 position - acording to thoese tiles. say, each tile is littlebit out of the position. due to perspective etc. so in that case - each pixel is not corresponding to each(corresponding) tile. right? if a certain tile of that mosaic is not adjusted ideally on coresponding pixel in the CCD matrix it means that it must be shown on at least two adjacent pixels. right? so, what we will see? which pixel/tile? another one: as you said: >>>> Only 480 lines are meant to be visible. (well, a few more actually but in NTSC DV, it is 480) PAL has a similar ratio of total to visible scan lines.<<<<< if i understood your point good - this means that those famous 625 lines are actually 576 lines/pixels in pal? yes? thank you for the answers. filip |
April 10th, 2004, 03:55 PM | #4 |
Major Player
Join Date: Apr 2003
Location: St. Louis, MO
Posts: 581
|
At the proper distance from the tile, and the size of the tile, also ignoring perspective, there would be a one to one relationship between tiles and pixels.
If, due to perspective or misalignment, the tile could be captured by two or more pixels. A tile could be missed altogether if it falls between ccd elements. If there are 625 PAL lines, the number of displayable lines will be less. Like ntsc, which has 525 lines, about 45 of those lines are used for vertical sync, retrace blanking and data. The same is true for PAL but I don't recall the timing. Each horizontal line also has sync pulses, horizontal blanking and color burst. 720 pixels on a line fit between the blanking intervals. |
April 10th, 2004, 05:55 PM | #5 |
Wrangler
Join Date: May 2002
Location: Vallejo, California
Posts: 4,049
|
<<<-- Originally posted by Filip Kovcin : mike, thank you for your answer, but i'm still confused.
so, to clear my mind - here are the questions: as i understand - if we are shooting a picture which consist in complicated patern, like mozaic or so, and by chance this mosaic is 720x576 tiles (say, in pal :))). Yes, because we have imposed that mosaic by using a camera. now, we are making compositionon our PAL system in wievfinder, and the whole masterpiece is in the frame. good. but! if we are shoting this mosaic masterpiece, we will not see the whole tiles correctly, because the camera is probably not adjusted in perfect "locked" - 1:1 position - acording to thoese tiles. say, each tile is littlebit out of the position. due to perspective etc. so in that case - each pixel is not corresponding to each(corresponding) tile. right? Absolutely correct. if a certain tile of that mosaic is not adjusted ideally on coresponding pixel in the CCD matrix it means that it must be shown on at least two adjacent pixels. right? Yes so, what we will see? which pixel/tile? Depends on a lot of things. How much of each pixel is subtended by that detail (understand that everything in the field of view is imaged on the CCD(s) but I understand you are talking about something like a picket fence. The algorithym used by the camera manufacturer for detecting detail, the amount of light reflected by the object, even the cleanliness of the lens and how well it focuses all colors and how still it is held will affect the image. Not to mention the quality of the focus and whether the object is single or multi-colored, e.g. red or white. Probably be reported as a half-intensity object by the two adjacent pixels. Except since a fence picket has length, it probably extends beyond those two pixels. Probably at an angle too. another one: as you said: >>>> Only 480 lines are meant to be visible. (well, a few more actually but in NTSC DV, it is 480) PAL has a similar ratio of total to visible scan lines.<<<<< if i understood your point good - this means that those famous 625 lines are actually 576 lines/pixels in pal? yes? Yes. thank you for the answers. filip -->>> The entire technology of detecting an image on a CCD and delivering to the recording media is very very difficult. Especially when there is a maximum design and manufacturing cost to be observed. Another slight problem is that
__________________
Mike Rehmus Hey, I can see the carrot at the end of the tunnel! |
April 10th, 2004, 08:53 PM | #6 |
Major Player
Join Date: Apr 2003
Location: St. Louis, MO
Posts: 581
|
When ya gotta go, ya gotta go. :)
|
April 11th, 2004, 07:26 PM | #7 |
Trustee
Join Date: Mar 2004
Location: Hilliard, Ohio
Posts: 1,193
|
Just got here so forgive me if we have all moved on but here is the next logical question...If we only use 720x480, what's the point of capturing more? In my case, the 3 CCDs of the PDX10 capture about 1 million pixels per chip. I understand processing in component but why would I go to all the trouble of making multi million pixel chips if the resultant image actually recorded to the tape in DV format (720x480) isn't going to use even a percentage of all those lovely dots?
Sean
__________________
‘I don’t know what I’m doing, and I’m shooting on D.V.’ - my hero - David Lynch http://www.DeepBlueEdit.com |
April 11th, 2004, 08:25 PM | #8 |
Wrangler
Join Date: May 2002
Location: Vallejo, California
Posts: 4,049
|
It takes a pixel from each CCD to make one finished pixel in the image. Red, Green, and Blue.
More pixels helps solve the problems discussed above with regard to picture details and CCD pixel density. The camera can make a better decision about where the detail should be placed in the visual field if it has more resolution than it can report to the tape. The downside is that small CCD pixels are not very sensitive.
__________________
Mike Rehmus Hey, I can see the carrot at the end of the tunnel! |
April 13th, 2004, 12:08 PM | #9 |
Major Player
Join Date: Jun 2003
Location: Miller Place, NY
Posts: 820
|
I believe the rationale behind having megapixel-plus CCDs in a video camera, Sean, is twofold; on the one hand, you have the still image capability, and on the other, in the case of the PDX10, you've got improved 16:9 recording. The chips in the PDX10 are sized at a 4:3 ratio, but the fact that they've got a million pixels means they can use a selection of them--that are arranged in a 16:9 shape--for widescreen recording.
Wish I could make more sense with that. I have a related question, I think; for the longest time, I heard people saying (with regard to this topic) that an analog TV signal technically doesn't really HAVE a horizontal resolution because, and I quote, "it's an analog waveform". Yet it bugged me, because when I get close to ANY television set, I can clearly see a division of pixels. There is a definite number of clusters of red, green, and blue dots making up the image on a given set, and given the time and motivation, one could most certainly count them all. Then I remembered they said "signal". This leads me to believe that the signal itself doesn't have a horizontal resolution, but the display device does...and that most likely, said number of horizontal dots varies from TV to TV, and manufacturer to manufacturer. Is this an accurate assumption? |
April 13th, 2004, 01:25 PM | #10 |
Major Player
Join Date: Apr 2003
Location: St. Louis, MO
Posts: 581
|
You're right. The display resolution and the signal "resolution" are not the same thing. You don't have to put your nose up to the screen to see the pixels. Take any magnifying glass, lens and maybe even some glasses and you can see the color triads.
The video signal is limited by the bandwidth of the transmission medium. There is about 60usec displayable on each horizontal line. The maximum bandwidth for ntsc is, uh, I forgot, 4.5 mhz? That's about 220 nsec. So 60usec/220nsec = 500 maximum dots per line (black and white only). Obviously my numbers are off but that's how you would calculate it. Color is limited by the bandwidth of the color burst and can only give about 180 per line(?). Sorry I'm not more exact. I'm really under the weather. |
April 13th, 2004, 08:15 PM | #11 |
Wrangler
Join Date: May 2002
Location: Vallejo, California
Posts: 4,049
|
Analog 'resolution' is measured by measuring the line-pairs. Every device has resolution values. That's why you see the test patterns that are used by broadcast stations have converging sets of line pairs. Those are used to measure the resolution of the system as it progresses through the station equipment and out onto the air.
See an earlier post in this thread about what line-pairs mean.
__________________
Mike Rehmus Hey, I can see the carrot at the end of the tunnel! |
| ||||||
|
Thread Tools | Search this Thread |
|