not happy with 1920x1080 p quality

G

gtimage

New member
Hi,

I have had my HVX200 for a while now and have been doing nothing but shooting and testing various scene file settings in 1080p.

Whilst I can see that this camera does a fine job at 720p - the results at 1080p seem to be no better than just upsizing the 720p results, and as these are only just good enough for 720 they look poor at 1080.

Soft and lacking detail - especially in wide shots. Any attempt to increase detail using the detail settings just results in unacceptable grain and doesn't really solve the detail problem anyway.

I guess I can get away with close shots where the perception of resolution is much easier to maintain - but once you zoom out the definition sucks.

The chip seems not to have sufficient pixels to provide a good image at 1080p.

Am I the only one who feels this way ?

Cheers

Geoff:(
 
Yeah, pretty much. :)

I shoot allot of my stuff in 720 just because I use the variable framerates so much.

I was out on a prairie yesterday filming the fog swirling around the prairie and shot both 720 and 1080.
There is an improvement in shooting 1080. I think Barry Green said it was about 20 or 30% better.

Some people have complained about the lack of detail for landscapes etc, but mine always have looked awesome.

I know when I play about anything on my Mac laptop it looks ok. It's only when I play it on my HDTV that I sit back and say Holy Smokes!
Well, actually I really don't say that but I'm sure the board would sensor what I really say. :)

Your idea of what is good is subjective. Everyone has an opinion on what is good.

How much detail do you want? I mean I shot some timelapse stuff of the moon and it was 238,712 miles away. That looked pretty sharp to me. :-)
 
hmm.. why dont u stick with 720p? also when u shoot this far off scenery.. how is the light? is there fog or generally wispyness from humidity in the atmosphere? I have shot distant scenery with a regular miniDV cam and have gotten quite nice detail considering the camera. But the key to this was shooting when the sunlight & general atmosphere was good.
 
I think there is more detail in 1080 but it is not sharp... it is more like soft detail if that makes sense. You can see there is more grass but the blades arent defined. The 1080 on the HVX is not near as sharp as many other HD cams but this is only really noticeable on wide detailed scenes and generally only when they are A/Bed.

ash =o)
 
I'd like to this guys work for better understanding.

The HVX is not as sharp as the Canon or JVC but, so what.

I wonder how much detail one needs ?

-

I have seen some Canon work, that, yeah, is almost overkill on clarity; like too sharp

-

It is subjective.
 
I'm quoteing Barry's tests a lot in this and what I've heard from others who worked with both cameras:

1. 1080p does offer a sharper image than uprezzed 720p. Different people argue different amounts but it's certainly there. Some say there is as much as a 25% resolution gain. I'd probably put it closer to 10-15% but it's still there. Try taking both 720p and 1080p and blowing them up a few times. You'll start to see the difference. 1080p downconverted to 720p also makes for cleaner/sharper 720p that incamera 720p.

2. The HVX200 has just as much detail as the JVC or the Sony (more detail than the sony if you are in it's version of 24 fps). There is a misperseption that the JVC is sharper since it uses far more edge enhancment. +7 in detail on the HVX200 is equivelant to -6 on the JVC. When it comes to actual detail, they are pretty much indisinguishible. The Canon XLH1 is sharper than the other three though. It has more detail rather than just extra edge enhacement.

3. A lot of getting your landscapes to look right has to do with messing with the settings. I did some landscapes I was initially frustrated looked very soft. I asked around what my issue was and people (mostly AshG) gave me some helpful solutions and I'm quite happy with the latest group of landscapes. Things like being willing to up your detail and turn down detail coring. This would produce a lot of noise if you were in low light but it's fine when you're shooting a landscape in daylight or even near sunrise/sunset. Making sure not to go above F9 is also apparently important for 1/3 CCDs. The point is that know what settings to use can produce drastically different results.

4.-Despite all I've said above, I actually agree that for 1080p, the HVX200 is not the best camera to use but niether are most others. The JVC HD100 can't shoot it at all and neither can the 65k Varicam. Of the smaller HD cams, the only one that I really think is truly viable as a 1080 camera is the Canon XLH1. But so what? Getting great 720p for under 10k is pretty damn sweet. I wish that it did 1080p better but considering the HD100 and Varicam don't do it at all and the Z1 doesn't do it any better, I don't see much of an issue.
 
Last edited:
The HVX starts out by scanning its chips into a 1920x1080 matrix. 720 is DOWNREZZED from that. It's not a case of 1080p being UPREZZED from 720, it's the other way around.

So in 1080 mode you're getting all the detail the camera can deliver. The chips are delivering something like 1400x810 (according to Panasonic's tech paper). The 1080 mode records 1280x1080, so it records just about everything the camera can deliver. The 720p mode scales it to 960x720 so it can't store all the detail the camera can deliver. That's why there's a resolution difference between 1080 and 720.

1080 mode (1280x1080) has twice as many pixels per frame as 720 mode (960x720). But that doesn't mean 1080 mode is twice as detailed! It's only about 15-20% more discernable resolution than 720 mode.
 
I understand that the calculation with pixel shift gives 150% gain in resolution.

960*540=1440*810 Sure there should be limitation how far you can push it, but if the ccds are offset with a half pixel vert/hor that would make it look like this.

PIXELSHIFT.GIF


That show that you would have for every normal ccd pixel 4 new uniqe pixels. There isnt a pixel thats the same from the other.

That gives us 960*540=1920*1080 (2X2)

With some sort of math you should be able to gain more resolution than just 1440*810, or is there something i missed?


The other thing I always think is funny is that so many complains that the HVX is soft. A great filmlook image is a soft but detailed picture. If you get more res but the same time get to sharp edges (checkboard edges) that totally destroys the filmllok for me.
 
I use shake to de-interlace. I have been shooting the camera built in1080 P mode and compared it to 1080 i which I subsequential de-interlace using Shake 4.1. The difference is just out of this world! Honestly ... I know you can't just retime everything ... but editing what I want in 1080i and then exporting from FCP into Shake and de-interlace it there ... the results are astonishing! I guess another reason to buy Shake 4.1 ... obvious that this formerly 3000 Euro/$ program does a better software job than whatever software is into the HVX ... So I guess I gambled right. Doing pre-production for a feature film ... I shot everything in 1080 i ... noticing me fields very much in Final Cut Pro 5.1 ... but now ... in and outputting it through Shake ... what a mindblowing result ! I compared this to 720P ... and gosh ... forget it ! If you got the means and system to film in 1080i and de-interlace it in Shake for film output ... forget all other options! ... Time consuming ... yes ... but I am talking about a final render of your edit ... I know my 90 minutes final edit will take 1350 minutes to render .. (ratio is 15) ... but that's only a night's rendering.
 
This strategy of deinterlacing 1080 with Shake, is this only useful when outputting to film? I plan to finish my doc on video for broadcast. I have lots of panoramic landscapes. I want them to look great. Can I get any advantage with Shake?

Tell me about deinterlacing as it relates to a video finish for broadcast. This seems like a contradiction to me.

If you get a great look, which happens to be deinterlaced, and you hand that off to a broadcaster and they broadcast it interlaced, is the Shake advantage retained or lost?
 
Id be also interested in the de-interlacing question: why would it be better recording in 1080i instead of 1080p, as the HVX got PROGRESSIVE chips???

I really cant understand where the effort comes from, as it would technicly mean:progressive chips-in camera processing to interlaced-de-interlacing in post..!

Id be really curious to hear about it..i am sure that you have done some tets before comparing 1080i and 1080p-so i believe you, but i am interested in understanding technicly why there would be an effort in this workflow...!
 
im not sure about the interlaced workflow, i always shoot progressive. and playback of 24 or 30p to 60i looks just fine.. just like any other film shown on TV. concerning the detail of 720 to 1080 though, it IS worth shooting 1080 if your going to be going to film out or something like that in my book. there IS some noticeable extra definition over 720 ..especially as barry said when blown up or inspected closely, and you can skip a cumbersome uprezzing step. i personally think that the 1080 mode produces some excellent imagery, and only use the 720 mode for variable frame rates ..alot, or longer straight record times when needed ..almost never.

a far as shooting landscapes, i think its generally the wider end of the lense that just softens the image. the DVX does the same thing, and so does almost evry other cam out there. Barry's advice is sage as always. additionally, (however marginally...) there is alot of refracted (and reflected)light in any scene, and nothing seems to generate more than a sunlit exterior scene. all that refraction tends to add a sort of "halo" effect around everything both to our own eyes, and to a camera. under that halo alot of edge detail , sharpness, and definition seem to fade away. one solution is to use a polarizer filter and you can dial out alot of that sunny glare off the landscape to bring some of these things back. another helpful thing is to lose the ND filters, and settle for a more closed iris instead. with a smaller iris arpeture, the light is forced to come through the lense and hit the sensor from a more defined point which can help reduce the "halo appearance", and cut off some of that refracted light. irising down wont make so much of a difference at a long range landcape scene (perhaps not even noticeable.. but still worth mentioning as a technique for getting rid of refraction) as it would say a close range close up of something where it can be much more noticeable since an iris is small aready, and for light to come from a good distance through it requires the angle to be pretty defined to begin with.
anyhow, lastly dont forget to focus. when the lense goes wide it appears that everything is in focus, but zooming in and getting a great focus could still improove the detail there pending your shot setup of course.
 
Additionally, keep in mind that setting your iris too small will LOSE definition and resolution. Shooting at f/8 is the smallest you should normally consider; maybe f/9.6 if you really need to but I'd recommend against it. F/11 is right out, as you could be giving up 30% of your potential resolution (according to a b&w res chart; if you count just the red spectrum you're probably losing even more!) So I don't necessarily agree with Neil about the aperture, I'd rather use an ND filter and keep the iris more open (around f/4) and use a mattebox to keep stray light to a minimum.

A mattebox is a very handy tool to keep stray light out of the lens. The HVX has a 16:9 antireflection plate installed in front of the lens to keep extraneous light down, and it's a nice idea but taking it further with a mattebox can only help.

And yes the lens does appear to be softer at the widest wide angles (just like the DVX, just like the PD150, just like a lot of cameras).

If I still lived in Vegas I'd try this: a stock HVX focused at infinity shooting a desert landscape at f/11 and 4.2mm, vs. that same HVX shooting at 6mm, using ND filters to get to F4, with a mattebox and a properly-rotated polarizer. I think the difference would be surprising to a few people.

It's overcast here so I can't get the same effect out here on the east coast; that harsh direct western sunlight causes great contrast and would let the polarizer really show off.
 
As far as the interlaced workflow -- please post some before & after pics; some native 1080/25p pics vs. some Shake-de-interlaced 1080/50i pics.

I ran some de-interlacing experiments in PhotoShop on a US model and the results were about what I expected: softer images than the 24p version. Sometimes not very much softer, sometimes significantly softer, but never was it better.

It'd be interesting to see what you're getting. Not so useful for us in the US since de-interlacing would give us only a simulated 30p, whereas in Belgium you're getting a simulated 25p.
 
HD Studio Belgium said:
I use shake to de-interlace. I have been shooting the camera built in1080 P mode and compared it to 1080 i which I subsequential de-interlace using Shake 4.1. The difference is just out of this world! Honestly ... I know you can't just retime everything ... but editing what I want in 1080i and then exporting from FCP into Shake and de-interlace it there ... the results are astonishing! I guess another reason to buy Shake 4.1 ... obvious that this formerly 3000 Euro/$ program does a better software job than whatever software is into the HVX ... So I guess I gambled right. Doing pre-production for a feature film ... I shot everything in 1080 i ... noticing me fields very much in Final Cut Pro 5.1 ... but now ... in and outputting it through Shake ... what a mindblowing result ! I compared this to 720P ... and gosh ... forget it ! If you got the means and system to film in 1080i and de-interlace it in Shake for film output ... forget all other options! ... Time consuming ... yes ... but I am talking about a final render of your edit ... I know my 90 minutes final edit will take 1350 minutes to render .. (ratio is 15) ... but that's only a night's rendering.
Click to expand...
Are you saying 1080i is better than 1080p*?

Do you stand it?


* EDIT -- Using HVX, of course.​
 
Last edited:
Just a note, however this Shake de-interlacing discussion turns out: Compressor 2 has the same optical flow feature as Shake 4, and will do the same high-quality conforming, de-interlacing, slow-mo, and re-sizing. The slow-mo is a multi-step process, but everything else is available right from the settings palette.

I don't know about Shake, but Compressor has a range of options affecting the quality of the output. Everything from "make it quick and it's OK if it looks bad" to "please monopolize my computer for the next week".
 
In a somewhat related note...how come many of the magazine reviews and comparisons that mention the HVX have it listed as a 1080i camera and several have said that it's not REALLY 1080p... is this based on anything? I have no links to present, I'm sorry :(.
 
ninjaleprechaun said:
In a somewhat related note...how come many of the magazine reviews and comparisons that mention the HVX have it listed as a 1080i camera and several have said that it's not REALLY 1080p... is this based on anything? I have no links to present, I'm sorry :(.
Click to expand...

It's a complete misunderstanding. The HVX is a progressive-only camera that doesn't even have the ability to scan its chips interlaced. It makes "fake interlace" out of progressive footage, not the other way around.
 
Hah! FAke interlace...... Take that Sony!

AS a DP i must say, im not bothered by its Resolution. What ever it is. I am more concerned with Dynamic range and Color accuracy.

Its plenty sharp and the footage ive seen for it looked nice and clean. I didint really see any more noise tha nany other cam. They all kinda looked the same actually. Infact i kinda liked the JVC's picture the best.
 
You must log in or register to reply here.
Back
Top