24p omission explained

[shaocaholica]

Amazing how someone says an unsubstantiated statement that has no factual merit to date, and this becomes fact. We don't know exactly what and how that CMOS intergeneration with the Digic chip set is working
.

Wasn't the original rumor that the 24fps implementation was a hack? If its a hack then they obviously didn't plan to have it. If they planned to have it and didn't implement it on day 1 is pretty bad too.

However, if they do pull it off without any glaring artifacts then I applaud them. I'm sure they are also kicking themselves for not doing it from the start if the rumors are true.

 

[PaulM]

Don't just underclock it, Canon!

Don't just underclock it, Canon!

I really hope they don't implement 24p by just slowing down master clock, unless it doesn't affect the CMOS read/reset speed. The CMOS readout is already a painfully slow 1/40s per 1080p frame. This has been established and corroborated by several users at cinema5d and dvinfo, Pango for one.

I'm trying to come up with an LCD shutter that will block light from the sensor for all but the period in which all pixels are simultaneously integrating. In 1080 30p mode, this period is just 1/120s per frame. The calculation is just (maximum exposure time per frame)-(read time per frame)=(maximum globally shuttered exposure). So with the same 1/40s read speed, 1080 24p could have a globally shuttered maximum exposure time of (1/24s)-(1/40s)=1/60s.

1/60 ain't too bad. If you shot 2.40:1 1920x800 24p, your read time per frame would be reduced to 1/54s. This would mean your maximum globally shuttered exposure time would be 1/43s. So Cinemascope 1920x800 24p with a 1/48 shutter would be totally doable. Even when factoring in the 1-3 milliseconds it takes to switch a high speed LCD shutter from an on to off state or vice versa.

Unless, in the process of implementing 24p, they slow down the CMOS read speed.