1080i or 1080p? Which one is best?

A church tech emailed me the following question. I asked his permission to create a controlbooth.tv post from my answer. He wrote:

Hey buddy, had a question for you. We are running our system at 1080i 59.94. We are using a Ross Black Carbonite switcher, bmd router, Nova Star processors for our Imag screens and LED wall onstage. We use Living as One to stream to our web site, also to FB and YT. Our creative department is asking us if we can convert all our stuff to 1080p, since they do all their videos in p. We normally just say remember that you need to give us stuff back in 1080i. But when we give them footage that we shot and they edit and forget about us being in 1080i it shows some of the interlacing. So, question is, is there any benefit for us to convert our system to 1080p in your professional opinion? I don’t like the idea of going to 1080i 59.94 because it shortens the distance of our SDI runs and don’t really want to spend a bunch of money reclocking all our camera lines and stage lines. Plus it would basically double our file size for recordings as well. I don’t know if it would affect uploading to Living as One. Going to 1080p at 29.97 I wouldn’t have to worry about losing distance on SDI. File size should be about the same. I will have to check all my sources and outputs to make sure everything will take it. Might have to buy a few converters, but no big deal. Is there really any benefit to going from 1080i @ 59.94 to 1080p @ 29.97?? Also, just to clarify, we have issues when we give them footage we shot in the worship center in 1080i. They take it to final edit and add a few last minute lower thirds, etc. We play it back on our system, again in 1080i and we will occasionally see some interlacing pop up in the lower thirds on fine lines. The stuff they put out to the web is great and no problems. Thanks for your time!!

Thank you for your question. There is a lot there to digest and some of the challenges you are experiencing may come down to unique factors (your particular church’s work flow, your equipment choices, staff/volunteer personalities, etc.), but I will try and answer the question “is there any benefit for us to convert our system to 1080p?” regardless to the best of my ability.

To set a technical foundation for my answer it is important to understand (if the folks at your church do not already) that there are various ingredients or parts which influence the perception of resolution and therefore the perceived quality of a video signal. These basic ingredients include (but are not limited to) spatial resolution, temporal resolution and bandwidth as expressed as a bit rate.

Spatial resolution is essentially the number of pixels visible horizontally and vertically. When we talk about “1920×1080” we are describing spatial resolution. However spatial resolution is only part of the equation. 1920×1080 on its own has no capacity to display motion or smoothness of motion.

Temporal resolution describes how often per second spatial resolution is updated. Thus, 23.98 means the entire horizontal and vertical spatial image is updated 23.98 times a second and 29.97 means 29.97 times per second regardless of whether the scanning system is “p” (progressive), or “PsF” (progressive segmented frame).

All of these, 1920×1080/23.98 (p or PsF), 1920×1080/29.97 (p or PsF) are 1.5 Gbit/s bandwidth SMPTE 292 signals, meaning when we combine the spatial and temporal resolutions a 1.5 Gbit/s pipe is required to transmit the image from point A to B.

The standards committees threw us a bit of a wrench however regarding the nomenclature 59.94 (or instead “50” for our European friends). A temporal resolution of 59.94 (or 50) can be found either in a “p” (progressive) or an “i” (interlaced) scanning system – and the differences between the two are substantial!

A 1920×1080/59.94p (or 50p) SMPTE 372 (or SMPTE 424) spatial/temporal resolution signal will not fit in a 1.5 Gbit/s pipe. That makes sense after all as the temporal component is twice 29.97 (or 2.5x 23.98) what it was previously. There are potential advantages to 59.94p versus 29.97p. For example content with a great deal of motion – for example sports – will benefit from the increase in temporal updates. Motion will appear smoother and less blurry without necessarily increasing spatial resolution to achieve that clarity.

The standards committees then created a compromise in an attempt to increase the perception of temporal resolution while maintaining the 1.5 Gbit/s SMPTE 292 bandwidth limit. Using the starting point of 1920×1080/29.97 and by scanning 540 of the 1080 vertical lines first followed by the other 540 of the 1080 lines second – all within the same 1/29.97 second of time – they created the perception of 1/59.94 motion refresh while still conforming to SMPTE 292 1.5 Gbit/s limit. Thus “1920×1080/59.94i” was born and became (for most) video broadcasters the de facto technical standard used (with notable exceptions being ABC, Disney, Fox and ESPN who opted instead for 1280×720/59.94p, which also conforms to SMPTE 292 1.5 Gbit/s).

Theoretically, adopters of a 1920×1080/29.97p (or PsF) standard would likely value spatial resolution over temporal resolution. Adopters of the 1280×720/59.94p standard may value temporal resolution over spatial. Most chose 1920×1080/59.94i, a compromise maintaining the full spatial resolution of 1920×1080/29.97p yet improving the perception of temporal resolution. All three of these standards (1920×1080/29.97p or PsF, 1280×720/59.94p and 1920×1080/59.94i) comply with the SMPTE 292.

While 1920×1080/59.94i is a widely adopted technical standard which offers the perception of a high temporal resolution while still conforming to a 1.5 Gbit/s pipe, it is not perfect. Images when still-framed can potentially exhibit saw-toothed interlaced artifacts. This was not an compelling problem in broadcast television since most images were/are shown while in motion (not frozen). However once internet viewing became popular and people attempted to grab stills from video content, some complained about the inherent interlace artifacts when scrutinizing the still frame quality of what was only intended to be viewed for a mere 0.016683 of a second.

The other issue which surfaced is while interlace is a common broadcast standard all internet streaming and video-on-demand (VOD) occurs progressively. That means that all internet delivered content which originates as interlace must be converted before being viewed via the internet, whether the conversion is handled by the content creator themselves, or by the internet content distribution network (CDN).

Poor quality internet encoders often utilize poor quality “i” to “p” conversions, causing interlace artifacts to occur in a livestream or VOD. Conversely, high quality encoders generally result in high quality motion images due in part to high quality “i” to “p” conversions.

Here is an example of content which aired on MSNBC, livestreamed to multiple internet CDNs and available on the internet as a VOD. It originated as 1920×1080/59.94i but displayed via the internet as progressive without objectionable interlace artifacts despite its interlaced origin.

Elemental is one of the companies who manufacturers high quality hardware encoders often used on professional streaming/VOD broadcasts.

So, should you convert a multi-camera live Imag & capture system to 1920×1080/59.94p? What would be the benefit in doing so? What would be the downside?

Unlike single-camera field capture with a post-production edit, conforming to a common technical standard when designing and operating a multi-camera production system is of paramount importance. It seems this may be the point on which there is a misunderstanding from your creative department. Adjusting a single-camera acquisition & edit from one technical standard to another is generally a very simple ESTHETIC decision. “I like the way it looks!” Whereas adjusting a multi-camera acquisition system from one technical standard to another is a substantial and complex ENGINEERING challenge. All items in the production signal chain: from camera heads, to signal converters, transport adapters, switcher, routers, monitors, recorders, encoders, cable-length decisions and ratings, etc., MUST conform to a single common denominator resolution/frame rate/scanning system standard. Achieving SMPTE 372 (424) 1920×1080/59.94p 3 Gbit/s may not be as easy as throwing a few menu switches. It is likely not all items currently being used are capable of 3G. If some equipment items will need to be replaced to conform to 3G SMPTE 372 production, might those monies be better spent on some other alternative technical solution?

The second potential issue is even if you can get your entire multi-camera system to operate correctly in 3G, 1920×1080/59.94p creates substantially larger record files, storage needs, longer upload times, etc., than SMPTE 292 1.5 Gbit/s would. Additional storage, wait time for files to transfer, etc., should be soberly considered when considering this change.

Lastly, is a 59.94p (“sports”) motion aesthetic an actual requirement from your church’s leadership? Acquiring in “p” or “PsF” (for example to avoid interlace artifacts caused by low quality encoding) can still be accomplished at lower frame rates such as 23.98 or 29.97 while still conforming to SMPTE 292 1.5 Gbit/s. Unless you require the increase of temporal resolution (e.g.: your church service has confetti cannons which go off every time the offertory financial goals are met – and 29.97 just doesn’t capture celebratory confetti well enough!), 59.94p may actually be solving your creative team’s problem with a temporal sledge hammer when other less taxing frame rate options may work equally well based on your content.

It seems the following may be prudent:

(1) All content, whether captured live multi-camera, or captured in the field and edited, should conform to the same technical standard, including frame rate and scanning system. As with all technical standards, they are compromises often resulting in choosing the lowest common denominator. If that in your case is SMPTE 292 1.5 Gbit/s, then all content (even from the creative team) should conform.

(2) The statement above is NOT to imply that the creative team must necessarily conform to you. 1920×1080/23.98p (or PsF) or 1920×1080/29.97p (or PsF) may be the happy medium all parties can agree upon.

(3) Even after all these steps are done, it sounds like attention needs to be focused on your encoding (and therefore “i” to “p” conversion) process. Artifact free conversions (like the Global Citizen example above) should be possible even if you do not invest in the highest quality most expensive Elemental encoder. What we can safely surmise is whatever you and the creative team are currently doing for encoding is not achieving the quality you both desire. This should be addressed as an encoding challenge and solution, not as a “i” vs “p” debate.

Photo credit: Nogwater on Flickr