I have a confession to make. I was over enthusiastic in my Interlace + HEVC blog post a little while ago. I confidently stated: “One of the great advances in the HEVC specification is that it does not support interlace”. This is blatantly untrue. I used a slightly more accurate statement in an earlier post.
“It [HEVC] has been published without interlace modes”
In truth, HEVC does support interlace because an H.264 stream is essentially a legal H.265 stream and H.264 as we all know supports interlace. So why am I making a big public confession? In truth, it’s because these two blog posts have created more heated responses that any others that I have written. Why is that?
I’ll cover many of the technicalities of HEVC over the coming months and in some new episodes of Bruce’s Shorts that are in preparation. For now, let’s look at the nature of the responses that I’ve received and see how they responded to my controversial statements about “Interlace is Evil and HEVC will help it die”
Heated “You got it wrong – please apologise” camp
The essential arguments here are that “we’ve got years of interlace content in our archives” and “we have millions of dollars of interlace plant” and “we can’t economically just turn it off”. I spend most of my life helping customers in this camp get to grips with the difficulties of mixed frame rate, mixed interlace / progressive, mixed film / video workflows. My opinion is that the core question here is not a technical one, but a commercial one. Namely “How do I get the best returns on the investment (ROI) I made in creating / obtaining this interlaced content”. Factored into this equation is not“should I de-interlace to progressive”, but “when should I de-interlace to progressive”. Doing nothing in the factory means that the de-interlace will be done in the consumer’s viewing device. The results are therefore neither under the control of the content owner nor the distribution channel provider. There is also not a single answer to this question because optimising the ROI is different depending on the value of the content, the scale of the organisation and the relative commercial returns from traditional interlace TV channels and progressive OTT delivery.
Heated “You got it right – hammer those nails in interlace’s coffin” camp
The essential argument here is that we need revolution and not evolution. The media world has evolved slowly to a crazy point with is discord of formats, frame rates, resolutions and scanning formats. Just now, in 2014, there is a chance to reduce some of the madness. Now, the wise words of xkcd should be screaming loudly at this point and we know that standards can make things worse as well as better … BUT … there are voices that point out that a new 4k integer frame rate, non-drop frame, non-interlaced clean standard might get us off the backwards compatibility millstone and into a future where we are linearly adding new problems to the industry rather than multiplying all the old problems by the number of new ones. If the cameras are progressive, the screens are progressive, the customer likes clean progressive pictures, then surely we (as an industry) can design some clever boxes that get the interlaced content profitably onto those screens in a way that works for all?
Here’s a proposal
AmberFin is now a Dalet company, but one thing has not changed – we like to do things right and we don’t mind if people are watching. HEVC did not add any new modes to support interlace, but you can use it to code interlace – just as you can use a standard that doesn’t know about moving images such as JPEG2000 to code interlace. If the ultimate goal is to get the best pictures onto new wide gamut TVs, tablets, phones, laptops, google glass and other devices, maybe we should conduct some industry experiments? If anyone reading this would like to help form the questions for those experiments then we might be able to raise an interest group to perform them.
My personal favorite questions to which I don’t know the answer would be:
- “Is frame repetition & scaling sufficient to get from HD 59.94fps to UHDTV1 119.88fps?” and
- “is the global workflow cost of fixing drop / non-drop timecode problems greater than the global workflow cost of installing integer frame rate upconverters for 120fps workflow”
There are many more questions. If you have some good ones, then I would love to hear them AND … I would like your permission to put them in a future blog post to see if other readers know the answers or share he concerns.
You may like to hear about Dalet / AmberFin’s latest news while you’re thinking. Why not sign up to our Post-NAB webinar?
Until next time.