Short Answer: There are two terminologies called "fields" and "frames". There are two fields (with unique, separate images) in one frame. When it comes to human eyes, the actual temporal resolution is always the "fields" per second, and never the "frames" per second of the analog transport layer (aka NTSC, PAL, SECAM). For NTSC, 60 fields per second, which means it is 60 effective frames per second ("fields") seen by human eyes, even with 480i interlaced live NTSC broadcasts. There are two unique images (fields) in one frame, captured at temporally different times, and displayed at temporally different times, creating an effective 60 unique images per second to eyes.
Long Answer: Read Below
Blur Busters Mythbusters Answer: FALSE"480i NTSC is always only 30 frames per second seen by eyes"
-- variants of this claim said by many people over the years
Reason: Semi-misleading legacy terminological technicality on "frames" vs "fields"
I even have TestUFO proof, see-for-yourself: www.testufo.com/interlace
Compare testufo.com in left browser window with testufo.com/interlace in right browser window. The motion clarity of 60fps progressive and 60fps interlaced is identical for horizontal motion. This is how analog sports broadcasts worked.
Specifically for interlaced material, "Frames" is NOT the number of images per second the human eyes get! It's the number of "Fields" per second, not the "Frames" per second, when you're measuring human visible temporal resolution of interlaced.
Don't blame yourself, it's a common confusion / misconception because of terminology.
Short Answer: There is still 60 completely unique pictures (fields) per second embedded in the 30 frames per second.
Long Answer: Even though terminologically it was 30 frames per second, there are two temporally-different images interlaced in the frame, like two very different images (fields) embedded in one frame (The two fields can be different. The picture in odd scanlines and even scanlines can be COMPLETELY DIFFERENT, and that often happens during sports broadcasts).
Fields are the actual "frames per second" seen by human eyes. The name "frame" in interlaced is just a confusing textbook vestigal written terminology not necessarily related to the number of actual pictures received by the human eyes per second. Do not confuse the two when it comes to interlacing!
Source Camera:
- T+0/60sec = camera capture even scanlines of scenery at this instant
- T+1/60sec = camera capture odd scanlines of scenery at this instant
- T+2/60sec = camera capture even scanlines of scenery at this instant
- T+3/60sec = camera capture odd scanlines of scenery at this instant
- T+4/60sec = camera capture even scanlines of scenery at this instant
- T+5/60sec = camera capture odd scanlines of scenery at this instant
- T+0/60sec = television displays even scanlines
- T+1/60sec = television displays odd scanlines
- T+2/60sec = television displays even scanlines
- T+3/60sec = television displays odd scanlines
- T+4/60sec = television displays even scanlines
- T+5/60sec = television displays odd scanlines
As inventor of TestUFO myself, I correctly simulate interlaced's full temporal resolution concept via the top UFO at www.testufo.com/interlaced if you do not believe me, and wish to see-for-yourself.
That is why 60fps sports was 2.5x smoother than movies, especially in horizontal motions in things like football, hockey, soccer, etc.
You did halve vertical resolution during vertical motion especially if the camera speed was an odd integer multipler of the interlacing speed (i.e. camera moving upwards/downwards 1,3,5,7,9 scanlines per 1/60sec), because it was a kind of a vertical picket fence effect blocking half of the underlying motion.
And yes, fade simulators could also simultaneously do interlaced simulation too, in a stacked fashion.
Theoretically, an infinite-refresh display can emulate the artifacts of all displays known to humankind. At certain refresh rates I can now accurately simulate DLP rainbow effect on a gaming LCD monitor now -- I even have a TestUFO Rainbow Effect demo (WARNING: Not for epileptics. Best for 120Hz+ displays). Try the Stars image, and roll your eyes to see the rainbow effect similar to a DLP of same colorwheel Hz. Virtually identical rainbowing to a DLP of equal colorwheel Hz as the LCD Hz. So use a 360Hz LCD to simulate a 360Hz color wheel. As refresh rates go up, it becomes easier to simulate certain elements of legacy displays of various kinds.
Yes, yes, yes, later technology captured 30 frames per second by capturing two fields and merging them into one frame. This is common in the streaming era now, and lower-end IPTV offerings (barf). Which looks less smooth than it looked on a CRT tube because the temporal resolution is definitely halved. You could have instead used a video deinterlacer and recorded a 60fps progressive video file to look 2x as smoother when streaming. But before the digital era (~1990), everything broadcast live was a temporal resolution of 50 or 60 pictures (fields) per second.
Just don't confuse what was done in the analog interlaced era. It was always a temporal resolution of 50 or 60 if it was broadcast analog live or if it was an analog video camera recorded directly to analog videotape. Back then, untouched-by-digital-processing 100% analog sports, sitcoms, weather, news, were all always 50 or 60 unique images per second hitting human eye balls for NTSC, for PAL, and for SECAM.
When digital video arrived, things became somewhat different and more flexible depending on what the camera operator wanted to do. A few shows such as Max Headroom used 30 images per second (half temporal resolution of a normal sitcom or sports recording), for artistic reasons, thanks to the advent of digital availability. But CNN/Fox/NBC/ESPN/WWF/etc continued to permanently use 60 fields per second as 60 unique images per second, all the way through the analog shutdown. As did many major sitcoms such as the original analog-video-recorded episodes of Sienfield (rather than telecined from film, or rebroadcast-from-film).
Digital reruns and digital rebroadcasts of analog videotape are often 30 images per second only because they were lazy and did not bother to deinterlace at the full field temporal resolution. Do not confuse this with how beautiful the same videotape looks like on my LCD TV, when I pipe it through a video processor first.
I know this because some video processors reduce temporal resolution in the art of converting the original 60-images-per-second analog material for display on a progressive scan display. So things became half as smooth as they would on a better deinterlacer. Many digital streaming services (YouTube, Netflix, etc) are lazy and lost half of the temporal resolution built into the videotape when it could have instead been saved.
When re-interlacing (playing back this digitally deinterlaced material back to an analog TV to re-interlace it)... The motion is no longer as smooth as it is when played by original videotape. Re-interlaceability could have been 100% preserved if deinterlaced correctly. That's a lot to cry about, when certain preservationists accidentally throw away some (or all) of the original temporal resolution by choosing the wrong deinterlacing algorithm.
If you were born after 1990s, it is easy to confuse the lack of 60-temporals-per-second television material, with the fault of video preservationists using the wrong deinterlace processing that destroys the original temporal resolution of the original interlaced material in the original recordings. Those who were born in the 1990s often make mis-assumptions about exactly how much temporal resolution were originally available in the original recordings, that could have been successfully preserved in this digital era.
You do realize I used to work for RUNCO, Key Digital, and TAW? And I worked with the Faroudja chip in the Immersive HOLO3DGRAPH. See my Linkedin, and scroll to older history ("See all 25 experiences" -> scroll to year 2000-2004).
You do realize I invented the world's first open-source 3:2 pulldown deinterlacer in dScaler too, more than 20 years ago? (Internet Archive with my name)
So don't tell me I don't know this bleep.
More advanced reading:
Interlaced was chosen because you could increase resolution without increasing bandwidth.
For NTSC, full 483-visible resolution at full 60 temporal was almost always done in the analog live-broadcast era. (Near the wartime era, before 480 was chosen for digital convenience, it was actually 483 scanlines with 42 scanlines of VBI at 21 scanlines per field). The analog electronics mechanism for interlacing necessitated an odd number to force the analog electronics to vertically offset the scanlines between the previous refresh cycles' scanlines. Digital used 480 as a round number for convenience.
525 was chosen because it is 3x5x5x7 which was easy to do with vaccuum tubes, and 21 VBI was chosen because it's 3x7. Prime numbers were a common system used to create the numbers of old analog standards such as 405i (early TV), 525i (NTSC), 625i (PAL), and 1125i (1980 analog HD), because it was easier to use stages of analog electronics to count the number of scanlines to sync to, and then slew the VBI (roll the VHOLD automatically) to lock the picture in place using 100% analog electronics. Conveniently, multiplying primes together results in odd numbers, which was a natural fit for the invention of interlacing.
Interestingly, the sizes of VBIs are also products of prime numbers (since it was easy for simple dumb analog electronics to "count" the number of scanlines per refresh cycles). NTSC VBI is 21 lines (7x3) and HDTV VBI is 45 lines (5x3x3). Eventually this became less important as televisions had a little bit of error tolerance (a few percent), allowing the odd 60Hz->59.94Hz (addition of color signal) or the 240p computer mode (interlacing at 524i instead of 525i caused the scanlines to overlap instead of offset, even on a 1950s TV), or the addition of closed captioning (digital data in Line 21 of VBI).
Now, enough of analog history. Let's talk effective temporal resolution of interlaced.
You do lose resolution during vertical motion because of how the motion interacts with the vertical interlacing . But interlacing is just metaphorically/effectively a vertical picket fence effect, that does not affect the original 60 pictures per second captured by the camera, despite embedding two pictures in one frame. Horizontal motion resolution was 60 unique pictures per second by eye in motion resolution. I even have TestUFO Interlacing in Vertical Motion, just play with even-and-odd pixel steps.
Try TestUFO Interlace #1 and TestUFO Interalce #2 (change your refresh rate to either 60Hz or 120Hz before trying these two tests for this specific demonstration) and you will observe one of them preserves full spatial resolution during vertical motion, because it's going at a motion speed at an even-numbered pixel steps per second, so it's alternating between the even and odd fields, preserving full spatial resolution. But even though the vertical spatial resolution is halved, you observe that the temporal resolution is still fully preserved regardless. So you could still be getting 60 unique images per second of half-vertical-resolution, don't confuse halving of spatials with halving of temporals.
So as you see motion speeds vary in interlaced material, you sometimes see vertical resolution vary a bit -- as the camera vertical panning speed slightly varied back in the era. Most people did not notice, but I did, when sitting 12 inches from a 27 inch tube -- and watching how the vertical resolution varied somewhat. Like two humans moving at the same time on opposite sides of a picket fence. If the speed was sufficiently different, you still saw the whole object rather than half of the object. But if both was moving in sync, the other human remained half-obscured. But if the human moved slower or moved faster, it revealed more of the resolution of the human at the other side of the picket fence. Now imagine interlacing as simply a vertical version of a picket fence. The underlying images were still 60 images per second, but how the underlying motion scrolled relative to the "interlace scrolling" (1 scanline/refresh), determines how much of the vertical motion resolution is preserved or not.
It's plainly evident in the TestUFO links. Still don't believe your eyes? Point a high speed camera (240fps will do for 60Hz) to the TestUFO Interlace (scrolling at fast speeds like 960 pixels/sec) in macro-closeup and you will see the interlacing in action, and realize that there are 2 completely different images ("fields") embedded in 1 "interlaced frame". Study closely. It'll click eventually.
A tracking human eye plots the temporals along the axis of the eye tracking motion vector, creating contouring (plasma), rainbow effects (DLP), ghosting (LCD GtG), interlacing (combing), multiscan (sawtoothing), coronas (LCD excess overdrive), and other temporals. Sub-millisecond temporals are now visible because of the Vicious Cycle Effect
One reason Blur Busters creates unusual display tests nobody else did before, is that I'm one of the few individuals in the world to successfully emulate a display in my head even before it is prototyped. I invented testufo.com/eyetracking, testufo.com/persistence and testufo.com/ghosting (now a peer reviewed paper with NIST.gov, NOKIA, and Keltek) in my head before creating the tests.
I can tell you a portion of the predicted artifacts of a theoretical display, after given the specifications of how the display is refreshed, long before the display is built. The more details you tell me (backlight behavior, temporal color behavior, other subrefresh behaviors, pulse width modulation behavior, pixel response behavior), the more accurately I can emulate a display in my brain (Note: I only answer hypothetical questions in Blur Busters Forums Area 51 -- and sometimes I even create TestUFO tests to prove that I'm right. 50% of TestUFO tests were actually created this way to micdrop a lot of online debate). Some people have a math brillance, others have a photogenic memory, but my brain has the display brillance.
