Quantcast

What next to improve picture quality

Tarbat

Well-known Member
Liam @ Prog AV said:
As I understand it (I may not be 100%) VIDNC and VIDPP are still AUTO mode i.e. they are still trying to detect film, but with a slightly lesser bias.
I don't think they do try to detect film. If I set to VIDPP or VIDNC, then all material, including film on BBCHD for example, show a "V" flag. The manual states that "For HD there is an AUTO mode and two video modes", suggesting that using VIDNC or VIDPP will not detect FILM, and so lose all the benefit of FILM mode for film material. I may be wrong, but that's how it looks to me, so I'm sticking with AUTO.
 

StooMonster

Well-known Member
Tarbat said:
5. Or maybe this particular episode of American Chopper was "cleaner", allowing the Lumagen to better detect cadence changes.
The two "I, Robot Bike" episodes didn't have much combing at all ... typical! Other episodes have been almost constant throughout.

The "Davies Love Bike" episode (one following "I, Robot Bike") had two sequences that combed quite significantly: one where Senior gave Mikey a car (it stays locked in F mode for about three minutes) and another towards the end when they rode bikes together.

In addition, when it combs the picture has the "classic" slo-mo effect as well as pinpoint combing lines (on my 1080p display) on edges of motion. Interestingly if one pauses and then continues in one of these comb sequences, it drops straight back into V mode; it's as if it acquires a F lock and then cannot drop it.

Bear in mind I've seen this on other channels too, e.g the channel promo of 'National Geographic HD' where the chap wakes up in bed and goes downstairs to make a cup of coffee; also I will record 'Brainiacs' on Sky One HD tonight, because last week the title sequence combed like mad.

Note: I am running latest firmware (Production Release 080406); I am not stupid and running 576i output and complaining that it's combing (thanks!); and as mentioned before running VIDPP and VIDNC fixes the problem, but like Tarbat I hate the idea of having to change deinterlacing methods dependant on material type.

Liam @ Prog AV said:
all video material throws up the odd false detected film frame all the time so you can't be gauranteed 100% video detection - especially with 1080i where it's still relatively new to the mass market
The other two 1080i deinterlacers I am comparing to Lumagen (one of which is the new Pioneer HD chipset in the PDP-5000EX plasma) do not fail on video material that Lumagen does. BTW it's material recorded on Sky HD so I can run it through all three of them, via HDMI and component too. :)

All I am doing here is highlighting that Lumagen is not perfect :eek: but feedback will enable them to improve 1080i cadence detection. If we can find specific examples that constantly cause the Lumagen deinterlacer problems then all the better, as it will give them something to work with. :smashin:

Tarbat said:
2. Maybe the SkyHD box is causing the problem. I think we've already established that my SkyHD box handles 576 material differently to Stoomonster's, since mine outputs 576i whereas Stoomonster's outputs 576p. Maybe it handles 1080i material differently as well.
Yes, EDID data makes Sky HD handle 576 different. I have a number of different VPs and displays, and Sky HD outputs 576 differently depending on EDID settings: e.g. Sky HD outputs RGB to Lumagen and YPbPr to iScan VPx0. Also, if yours is not outputting 576p to Lumagen by default then it's different to all the other posts I've read on AVforums and DigitalSpy where everyone seems to have same 576 configuration as me. My Sky HD model is Thomson 607120 btw.

StooMonster
 

Tarbat

Well-known Member
StooMonster said:
The "Davies Love Bike" episode (one following "I, Robot Bike") had two sequences that combed quite significantly: one where Senior gave Mikey a car (it stays locked in F mode for about three minutes)
Okay, I've seen this, and yes, it stayed locked on F (wrongly), but only for 28 seconds. I wonder if Lumagen can do anything to tweak the HiDef film cadence detection routines? At the moment it appears that once F mode is detected it stays in that mode for too long, without checking the cadence, which I guess is good for watching film material.
 

NonPayingMember

Previously Liam @ Prog AV
StooMonster said:
The other two 1080i deinterlacers I am comparing to Lumagen (one of which is the new Pioneer HD chipset in the PDP-5000EX plasma) do not fail on video material that Lumagen does. BTW it's material recorded on Sky HD so I can run it through all three of them, via HDMI and component too. :)
Except for in the worst case where the processor stays in F too long and suffers combs, you wouldn't actually know if your plasma had flicked into F and back to V again if the material was solid for those few frames since the finished image would be exactly the same (I'm talking in the case of two fields of an interlaced image, that just happen to be indentical through a lack of any motion)....

This was what I meant. I did not mean to suggest that all other processors have combing errors as they don't. My point was aimed at people who have briefly flicked through this thread and will now go home, turn on Live Flags, then panic that the V and F is changing every now and then but not understand that this is normal!!
 
Top Bottom