PXW-X320 calibration/alignment settings

[Paul Anderegg]

Looking much better....here are your settings with addition of white and black limit at +35/+30. The door edges seem to have excessive white and black sized edging....would this be where I would adjust the FREQUENCY level to make them thinner? Any other suggestions based on this screen grab are greatly appreciated. Getting a little too close to SD uprezzed to HD look. :-)

[Christopher Young]

From my POV you are going in the right direction. I think our Mr. A Chapman sums it up pretty concisely in this parra.

"As well as adjusting the amount of detail correction (Detail Level), you can also adjust the ratio of horizontal and vertical correction, the maximum brightness or darkness of the applied edges (white and black limit). The thickness of the edges (frequency), the minimum contrast change that the correction will be applied to (crispening) and you can tell the camera not to apply detail correction to dark areas (level depend)."

It really is a matter of a combination of these above adjustments. They can vary considerably between SD and HD and even between 720 and 1080 not to mention 4K.

Detail Correction, Crispening, Aperture and Image Sharpness | XDCAM-USER.COM

Chris Young
CYV Productions
Sydney

[Paul Anderegg]

I think i really nailed the right combination of settings for detail...the still frames are with your hi sat matrix, and some custom detail settings i did pointing at crap around my room at 12db gain while fiddling with the menu. I am running +20 crispening, +70 for both black and white limit, and 0 for detail. The massively high limit values were the key to keeping the detail from looking cartoonish, and leaving all the other settings pretty much zed out kept my shadows from going all muddy. Cannot believe how horrible the PMW320 looked with stock gain settings!

Chris, let me know what looks odd to you or that I should pay closer attention to, and I will tweak a little more. Again, I did these in my room at night, not particularly pointing at any subject manner type I would actually record. :)

[Christopher Young]

Paul I see nothing really objectionable in any of the pictures. The extreme highlight spots, the white bits and floro reflective jackets, always bad news in hard direct camera light, all look sharp but 'smooth' sharp, not hard aliased edges which is often the case on high contrast transitions when detail is too high. Look if I had to work with that camera like I see it now I wouldn't have any qualms about it. Overall the image looks fine to me. There again you are the final decision maker. If you are happy with the results and more importantly if the client is then it's probably time to save those settings for that sort of work. Good results!

Chris Young
CYV Productions
Sydney

[Paul Anderegg]

I finally eliminated the horrid green/yellow cast on shadows...leveled the R G and B channels using the flare control on my Leader scope...yay!

R and B black levels were accurate, my white was accurate, and I see most everyone is throwing matrix corrections at Sony cameras to adjust this. Fujinon HS18x5.5BERM lens.

YouTube (Short URL)

[Andrew Smith]

What I've always liked is seeing the broadcast stories with Mr Anderegg's footage in it to appreciate the finished product as a whole. Just saying. ;-)

Andrew

[Paul Anderegg]

My DSC Camalign chart expires in 5 days, so came back here to see if there is any last minute scoping I might want to do.

Hey, Chris...my work doesn't allow me to shoot with my personal a7SIII which does amazing HLG3 4K 120p HEVC and handles highlights amazing. I am stuck working full time with the X320, I can't even bring my Z90 to work. :(

For content like this, how do you think HG would perform comparatively? I think this was with BBC 95%/+30 knee settings? I have always hated the color of flames on these 3 chip Sony cameras...with stock settings you can spot them on the news because the transition from white clip to orange glow is typically a shade of green instead of yellow! :-P

I played with HG once before, but didn't know what to do with it once I got it back in FCPX, not sure if there exists a combo of settings for HG live or night news.

Paul

YouTube (Short URL)

[Paul Anderegg]

As for the rest of the settings I finalized on a ways back, here is night VO shot by me and day VO shot by my dayside companion. She also has an X320 and I transferred my settings into her camera, so this is the best evaluative day and night for the settings I have been using. If you see anything that may warrant attention or adjustment, please feel free to chime in. Mostly wondering about things like the knee in day on faces.

YouTube (Short URL)

[Christopher Young]

Originally Posted by Paul Anderegg

For content like this, how do you think HG would perform comparatively? I think this was with BBC 95%/+30 knee settings? I have always hated the color of flames on these 3 chip Sony cameras...with stock settings you can spot them on the news because the transition from white clip to orange glow is typically a shade of green instead of yellow! :-P

Geez Paul. This is an old resurrection! Look, I don't have access to a PXW-X320 these days. Just dug through my old notes and these were some of the settings I used. You are welcome to them if of any help.

Various different knee/slope settings depending on what you were shooting and what you wanted to retain in highlight detail. The manual stated knee at 90 and slope at 0. The problem with those settings is that the higher the knee setting is the less compression headroom you give yourself to contain the highlights to obtain a half-decent looking roll-off. The lower the knee the less savage the slope gives you a more natural-looking highlight compression. The higher the knee and the harsher the slope the more off-colour artifacts you will see. When the R, G and B channels are compressed in the knee they do not all compress equally and evenly in a linear manner hence you can get those odd color casts. In bright cloudy skies sometimes you will see hints of pink and yellow under high knee compression.

A knee of 85 and a slope of +14 will give you in the order of 1.5 stops of headroom for highlight compression

A knee of 80 and a slope of -2 will give you in the order of 2.0 stops of headroom for highlight
compression

A knee of 75 and a slope of +10 will give you in the order of 2.5 stops of headroom for highlight
compression

If you have a lot of human skin tones in the image I would be careful of the 75 knee as that is getting close to Sony's default 75% skin tone levels. i.e. 75% zebra level. I nearly always kept my skin tones around 65, still do. Especially on the later cameras such as the s7Siii and the a7iv which I have been using lately.

Bear in mind your 709 gamma curve gives you a range of about six stops. If you can keep 2 stops (80 knee -2 slope) which is a third of your 709 dynamic range to handle the higher end exposure curve you get a much nicer looking image. Not such a harsh 'video' look.

The quickest way to check this is on a nice very bright cloudy overcast day expose an outdoor image to your satisfaction. Like a street scene with people in it. Exposing for them and their surroundings. Make sure you have a good slab of the hot overcast cloudy sky in the shot. Now without changing the aperture from your ideal subject exposure try all the above-suggested knee/slope settings. You will find with a 75 knee with a +10 slope you will retain a lot more highlight detail and get a much more natural-looking result.

Now, none of these above numbers is set in concrete but they are a pretty good starting point to work from. The basic settings relationship will give you varying highlight handling. The most common setting I used was the knee at 80 with the slope at -2. Normal blue sky sunny days I would go with the 85 knee and +14 slope. I used to keep three lots of identical PP settings in the camera except they had different knee/slope settings. Switching from one saved PP to another allowed me to quickly select the best knee/slope combo for the scene/s at hand. These settings will also work well under nighttime news conditions depending on how much light there is and how much detail you want to retain in overbright areas. Experiment.

Regarding the detail settings, I see I have noted them down as follows. Detail -10 to -15, Frequency +99 and Crispening -35. These I always find are a bit lens dependent. With the stock 16 x these figures worked quite well. I know Frequency at +99 sounds extreme but it works well with the detail settings down to -10-15. With the Crispening at minus values, you will raise image noise marginally. But at low gain settings I find it's fine. If using a lot of high gain, 9dB and up at night I would have a Hi-Gain PP with the Crispening at 0 or even up to +25. You just need to try these settings out to see what works best for you.

Sadly not a lot you can do with those savage blue LED type lights. You can minimise the harsh edge effect caused by the overdriven blue channel by setting a nighttime PP with a lower saturation setting in the blue channel. Do this by setting the camera into "Multi Matrix" mode not the default Color Correction mode. Then select the "B" axis and wind down the saturation. Leave the camera in Multi Matrix mode. I seem to recall using something like about -30 saturation in the B axis. It will help a bit. The three chippers seem to have more trouble in handling the high luminance high saturation of LEDS in their blue channel than single-chip cameras do. That's partly down to the demosaicing algorithms used in the Debayer process which is of course a process not used in three sensor cameras as they have full bandwidth on each sensor channel, R, G and B.

Stay safe now.

Chris Young

[Paul Anderegg]

Thank you for the detailed response, as usual!

The dayside SOTs in the sun, they have that hard clip to them, with no apparent knee crushing approaching the clip point. IMHO, I think that looks good compared to the strange facial tones and posterization you typically see in similar circumstances with lowered heavy knee applications. Most of my stations photogs just run faces up to 100% to make the image brighter and bring up the backgrounds, so exposure isn't something the newer generation of "learned to shot on an iphone" have experience with, so you see that compression a lot with them.

Still never found a completely satisfactory set for detail, will experiment further. I find my current settings look both soft and harsh at the same time. Our lenses are the $13k Fujinnon HS18's with extenders, which are absolute garbage below f2.4, and when I punch in to focus the peaking is disabled and I see them for what they are.

Last question, what distance do you recommend for setting backfocus? I find myself rebackfocusing in the field all the time, because I am never sure if it is my lens quality, detail settings, or the BF that is making my wides look soft. I typically switch to 1080p to set backfocus, as focus mag crops tighter in 1080 than it does in 720 making it easier to perform the adjustment using the viewfinder.

Paul

[Christopher Young]

Those Fujinon 18x HD lenses for the 1/2" cams were pretty awful. Especially when it came to CA problems. I had one and bailed out of it pretty quick. I got much better results in sharpness, axial tracking and CA performance when shooting HD using the last of the Fuji Digi SD broadcast lenses. They have no ALAC circuits but they were the very best glass Fujinon made. It is the same glass but with different coatings that went into their top range HD lenses the HA series.

With the top of the range SD glass they didn't have any electronic correction trickery to fall back on. To get the max quality out of them they had to use the absolute best lens grinds with the finest tolerances and totally hand assemble and align them. Now, this is not an old wives tale. It is straight from the horse's mouth. A senior lens designer at Fujinon said the following during a discussion with me regarding lens design and accuracy. I won't reveal his name as he is now very senior at Fujinon. In an email he sent to me he said:

"Hi Chris,
It's good to hear happy endings!
We want to sell more new HD products, but fortunately for many of our users, our very best top of the line SD lenses are still delivering extremely high-quality images on the bulk of HD cameras... Better in many cases than any of the lower range HD lenses other than our very best which is our HA series. The last of the best SD lenses and the first of the HA series used the best elements Fujinon had at the time."

Go figure!

This particular engineer told me the best way of setting up your back focus is to run the camera for a good half hour or so before you back focus. This way the camera gets to its full operating temperature range so all heat expansion in the body and lens mount has taken place. If you back focus on a stone-cold lens and camera assembly it can go out of whack as it heats up. I remember this as being a problem on the very first 2/3" Varicams from Panasonic. I thought there was a problem with the camera. I was actually shooting demo material for Panasonic for a trade show on one of the first production Varicams to arrive in Oz. This is when I spoke to Fujinon to try and find out why the lens was losing back focus

Back focusing. Once the camera is warm set up your back focus distance between 2 to 3 meters from a Siemens Star chart and adjust the back focus visually as you would normally do. Now to fine-tune the back focus use a WFM (waveform monitor) set to luminance only. That's the "Low Pass" selection. Still shooting the Siemens Star chart now gently rock-n-roll the back focus ring and keep a sharp eye on the waveform monitor. You will very quickly see that the waveform changes as you move the back focus ring. The back focus is optimal when the low pass waveform is at its "Null point." You will soon see what I mean. The waveform will change its pattern one way and then start to change the opposite way as you adjust. Find the middle point so that the waveform is neither moving one way nor the other. This is the Null point of the electronic signal. This is the most accurate way of adjusting the back focus. There is no ambiguity involved like "Is that sharp?" "Are my eyes tired?" It's called the circle of confusion for a reason while looking at the monitor. How good is the monitor is another question. The Null point of the electronic waveform, neither negative nor positive-going is is a measurement of the accuracy of the light focused on the camera's sensor. It's called aiming for the "circle of least confusion." I've been back focusing multiple cameras on OB's this way for years now and have not had any problems in cold, rain or shine.

Give it a try. :)

Chris Young

[Paul Anderegg]

Chris, maybe you can help suggest some changes to my detail settings. I thought I was well dialed in, and I am for 0db, the camera looks really nice and video smooth/sharp in 720p. Once I kick it up to 9db, things change, and there is a severe harshness aparent. Hoping you woudl be able to recognize which setting I may need to play with to real this in. Thinking maybe crispening or level depend, but you might have a better eye for this. Attaching a few 0db and 9db still frames, but basically the harshness is most visible in things like car headlights, chrome crinkled foil type of objects, and basic edges.

First 5 pics are 0db, last three are 9db.

Paul

*I miss my old Panasonic cameras ability to store completely independent detail values for each gain setting!!!!!

[Andrew Smith]

I absolutely love this sort of thread. Especially as we get to see more of Mr Anderson's work.

Andrew

[Paul Anderegg]

Was definitely the CRISPENING, was at +25, when dialed to -25, the issue was resolved. :-)

Paul

[Andrew Smith]

Whoops. Should have been "we get to see more of Mr Anderegg's work". My brain must have gone all Matrix on me whilst I was typing.

Andrew