Making Blu-ray Discs from EX1 Footage (Test Results)

basspig

basspig

Veteran
I published my second Blu-ray title this month, which contains some footage shot at -3dB, with a remapping from IRE 109 to IRE 100.

I did the editing in Premiere CS3 and the disc authoring in Encore CS3. (It was also my first successful motion menu).

I made my initial disc with no processing to the footage, other than the remapping to fit the highlights within Premiere's rendering engine dynamic range.

The result on the parade footage is stunning. There isn't any grain anywhere to be found in the picture, not in the shadows, not the mid tones, not the sky.

Initially, I made the Blu-ray using Media Encoder and chose the h.264 CODEC.

Now I was only seeing bitrates around 18-25mb/s on the Blu-ray player, a Sony BDP-S301. Wanting to compare with several Hollywood releases and megabucks cameras and lenses and encoding, I loaded up several movies and watched scenes that I found the most detailed, then switched back to the EX1 Blu-ray disc to compare quickly.

My overall concensus is that virtually every film I compared contained grain and most were lacking in very fine detail to the pixel level on my 47" full HD display. (I was sitting crouching just 1 foot from the monitor and really studying pixel by pixel here, pausing the video periodically and looking at grain, mosquito noise and other artifacts.)

The worst of the films was Full Metal Jacket. It seemed to have not much more detail than an upscaled DVD. Hart's War was little better, but contained halos around edges, which made it look worse to me. The best film I saw for clarity was "International", which had details that rivaled my non-sharpened XDCam footage.

I also compared "Over America" a NHK-produced video of aerial footage of major US cities. That appeared a tad better than looking at HDV footage. Detail was lacking, halos around edges, copious grain/noise everywhere. Colors subdued and washed out. Complex images with foliage were 'buzzy', indicating CODEC stressing. The bitrate on that Blu-ray was around 44mb/s.

Most of the commercial Blu-rays had data rates reaching 48.35mb/s on the display meter, while my Blu-ray was peaking at 25 and usually around 11-18. So I was getting similar picture quality, though a bit of mosquito noise visible in complex scenes when the disc was paused for close inspection (8" from the screen).

One of the notable things about my footage is the intense colors. The movies have subdued colors--particularly the reds. My Blu-ray disc has colors that are lively, dynamic and punchy. Now I realize that HiSat matrix won't go over a Composite video connection very well, so maybe that's why the movies are toned down in color saturation.

At that point, I wondered why Adobe doesn't allow the max bitrate to go all the way to 48mb/s (it stops at 35). But that got me thinking.. I did some test renders as follows:

I added a small amount of fine grain to the footage.
I added a 0.4 radius at 100 amount of Unsharp Mask.

I rendered this with the settings I used to make the BD disc. It looked pretty bad, with blockiness and the grain was not reproduced faithfully at all.

Next, I bumped the target bitrate from 25 to 35 (maximum), with the max left at 35 as well. Rendered the same clip. This clip, though 50% larger in file space, rendered the grain faithfully and there was nearly no mosquito noise despite the aggressive Unsharp Mask applied. The footage had that hyper-crisp look with the slight gritiness of film like I had seen in the "International" Blu-ray. It's simply gorgeous.

So this bears out that it's not the max bitrate alone that determines the amount of artifacts in the rendered video.

Last, I rendered another version of this clip, this time without the grain effect, just the Unsharp Mask. That render was nearly always free from any trace of mosquito noise. Red and white stripes on the flag were sharp and clear, despite that the camera was hand held and moving around constantly. If I had locked off on a tripod, the results could have been even cleaner. Pausing the video and peering at edge details, color boundaries and textures was a delight. I could not believe I was looking at recompressed footage.

My conclusion is that for really important footage that contains down to the pixel detail, push that encoder bitrate all the way up. Then you can even tolerate some post unsharp mask with a very small radius, to give that extra crisp fine detail.

The more I study the wide shots, the more impressed I am with the Fujinon lens. It's really sharp and at wide to mid zoom, chromatic aberrations are about nil. I've seen worse CA on Stanley Kubrick's films. It seems there is no reason why you couldn't shoot a major motion picture with the EX1/3 cameras. Carefully-handled, the footage can look better than 80% of the existing films on Blu-ray, and can match the best of the best. Imagine what you could do using the SDI output to a really high end encoder. The possibilities are endless.
 
Last edited:
Congratulations

Congratulations

I have achieved similar results with my EX3 and Encore CS4 for two Blu-ray projects that I completed last Christmas. I built custom motion menus for both projects, one 20 minutes (25Gb disc) in length and the second 120 minutes in length (50Gb disc). I stayed with Encore's Mpgeg encoder set to 35Mb/s so as to maintain the EX (HQ-35Mb/s) rate for 1080/60i. As you have articulated, the results are indeed stunning.

Some folks are claiming that their Blu-ray discs will not play on some of their customer's players. So far, I have not had this problem and the customers are blown away with the quality. I burn with the Sony BWU-200S. I did have some problems burning 50Gb Blu-ray discs reliably using Panasonic media. I think that Sony's 200S laser may come up short on power when burning to this media at X4 speeds. I have yet to try other Blu-ray disc media to confirm this.

Next month, I will be back in the field for some outdoor nature projects. These will afford me time to try some 24p and 30p footage. Cheers!
 
Barry, 'glad to hear other success stories with Blu-ray production.

The issue with players is, I'm becoming convinced, a political one. I read recently that Sharp, as a company policy, will not allow their players to read BD-R. The reason the company cites is "to thwart movie piracy".

Also, I had read that the Blu-ray spec requires AACS encryption, or the player SHOULD NOT read the disc. The players that DO read these discs are not 100% compliant with the spec.

In my own testing at Sears and Best Buy, only about half of the players tried will read my discs. Most rejected it with "cannot play this disc". My own Sony BDP-S301 will play ALL of my BD-R/RE discs--even when my computer can't recognize the disc it burned! I and a friend went shopping for his new Blu-ray player, and we surveyed a Samsung BD-P1600. It, too, played all of the five BD-R/RE samples I brought along with me. It's a profile 2.0 player with BD-Live.

This whole issue of cranking up the bitrate to improve picture quality is making me want to redo my demo reel again! It looks awesome, but that extra 15% picture clarity can make a difference. Comparing against the best movies is usually a sobering and depressing experience, but in this case, it was an uplifting one!
 
I added some small amount of very tiny radius (0.4) Unsharp Mask in Premiere CS3 and discovered that Unsharp Mask does not work properly on footage that exceeds 100 IRE. Even with the Levels set to 235 to restore highlight details, adding Unsharp Mask clamps the highlights anyway. I think it is being applied on the full range of signal, rather than being applied to the signal AFTER the Levels filter. Stacking order has no effect. So no unsharp mask, unless you don't mind losing highlights...
 
This problem goes deeper than Unsharp Mask. As it turns out, the Adobe Media Encoder is NOT capable of handling footage (even footage that's been rescaled to 0-100 IRE) where the original footage exceeds 100 IRE (super white). Apparently, even using the levels within Fast Color Corrector doesn't carry over to the Media Encoder.

My workaround is to render to uncompressed AVI (thankfully I have two RAID arrays in the editor, a 1TB and a 2TB array) with the levels adjusted so the clip is rendered video-legal levels. Then I import that into the Premiere project and apply unsharp mask and send to the Media Encoder. This way, no highlight data is lost. The downside may be quality loss due to an extra rendering step, but being it's to uncompressed 10-bit files, hopefully nothing was lost over going direct to h.264.
 
Firstly, IOU virtual beer for your sharing of your tests. Many thanks from the rest of us out in the community.

basspig said:
The issue with players is, I'm becoming convinced, a political one. I read recently that Sharp, as a company policy, will not allow their players to read BD-R. The reason the company cites is "to thwart movie piracy".
Click to expand...

Although there are echoes in the early days of DVD, I think there's going to be a cut off point when all the various levels of compliance get sorted. Just as The Matrix marked a watershed for DVD, so we'll get something that does the same for BluRay.

Are you following the ongoing story narrated by Bruce Nazarian? AACS-Free for low volume BluRay on BD-R should be in the specs, whatever Sharp's policy is. The industry hasn't quite got into it yet, but it's necessary to ignite the event and wedding markets which can be key to encourage consumers to update to HD.

The rumour mill is cranking out better licensing and an easing of low volume (sub 1k units) restrictions so you need NO paperwork for a certain level of job. All we need now is that STANDARD. Let the consumers shout at Sharp (and the rest) for building sub-standard players, rather than yell at the BluRay standard.

BTW, did you do any tests with 720p upscale? I did some consumer testing and was shocked at how few people couldn't see the difference between good upscaled SD (PAL) and HD, though my tests continue. (Though they're following a familar pattern where the general public want big bright sharp pictures that bounce around, and a little wafer thin layer of congnicenti want our carefully created imagery that can only be appreciated by turning most of the 'image enhancement' of their screens OFF. Sigh.)
 
Being that we're on different continents, I suppose a virtual beer will just fine. Much appreciate the good words!

With regard to fixing the Blu-ray compatibility issues, consumer awareness is #1, IMHO, because sales of players to consumers is what can be affected when it becomes common knowledge that certain players may be neutered as far as playback is concerned.

I did read some short article last month about the license fees being dropped from the five-figure range for a production run, to pennies per disc. I think it's a step in the right direction. But all of this will require new authoring software, or major updates to the software we have. I am hoping this will all get sorted out very soon.

I have some footage where I mixed 720P with 1080P because some of that footage was 60/24 overcranked. For me, knowing what to look for, I can tell the difference. But the 720P still looks better than the 1080i out of my V1U.

I think you're referring to DVD (480P) upscaled to 1080P. My Sony BDP-S301 does a better job than my Oppo DV 981HD. The Sony is crisp and sharp, while the Oppo produces a soft picture. I have a lot of SD DVD that I made from XDCam footage, and it looks darned close to broadcast 720P, because I add a lot of edge enhancement in post, just before I render to MPEG DVD. In fact, I push details right up to Nyquist limits, yielding a detailed picture that rivals 720P broadcast, like we typically see in big box stores here in the US. In this case, it blurs the line between DVD and so-called "720P".

I strongly believe that part of the sluggish adoption of HD in the US is due to improper demonstration setups in most of the stores I visited. Imagine this: a $6,000 65" Pioneer Kuro display, hooked up via composite video and displaying 480i! I found just that at Best Buy in Danbury, CT. I can just imagine what customers think as they walk by that --"This is HD? Gee, it doesn't look any better than what I'm used to at home."

Then there are stores that only hook up component, and run the TVs at 480P, some at 720P. Better, but still far from HD.

And finally, there are the movie sources themselves. There's a world of difference between "Full Metal Jacket" and "The International" in terms of picture detail on the Blu-ray. FMJ looks not a whole lot more detailed than upscaled DVD, whereas TI looks hyper-sharp. If the public is not being shown the best of the Blu-ray discs, on screens large enough to subtend a large enough field of vision where the viewer can resolve down to the pixel, it's not going to make a whole lotta' difference. And there are some people who still think VHS looks fine.

Frustrating as that is, I spend hours experimenting with different renders, various levels of Unsharp Mask, troubleshooting the weak links that loose picture dynamic range, etc., for my own benefit. I simply want to see how far the tech can be pushed, to achieve the best picture possible.

My latest processing recipe has modest amounts of unsharp mask added in post, with a sub-pixel radius value and I'm liking the result a lot better.

I plan to do some testing to see if in-camera sharpening or 'detail off' and post-sharpening really makes a significant difference in MPEG artifacts on the distribution Bly-ray. My theory is that if the XDCam encoder has less HF detail to deal with, overall quantizing error will be reduced. Thus, we would be starting with cleaner footage and THEN add all detail enhancement during the final render to h.264, thus putting burden only on the final render. Viewing "busy" scenes should make possible the comparison of smoothness to the textures and fine details in the two versions. Testing all of this requires actually rendering and authoring to BD-RE, since the way a set top player handles it looks different than the way a computer handles it.

What I've accomplished so far is gaining an extra 1/2 f-stop of highlight detail (and gaining back that 10% or so of the frame that was lost to white-clipping), pushing down everything else proportionately (thereby deeping saturation and lowering noise) and adding detail in post to bring edges quite literally to single-pixel boundaries of clarity. Pushing too far causes clapboard exteriors to 'buzz' or otherwise alias on wide shots where the detail is too small for the pixel resolution and eye to resolve. Striking a balance where this 'noise' is reduced, results in a smoother image. Getting it just right is where the lamp post a block away has edges that are 1-pixel wide, without halos.

I spent two hours last night with a EIA 1956 resolution chart and the EX1, observing how the various detail parameters affect sharpness. I reached a couple of conclusions:
Adding detail has zero effect on how much detail the camera can resolve.
High values of White Limiter skews the ratio of H & V sharpening away from H sharpening. I found that I had to dial in negative values of H&V sharpening ratio to balance it out.
Backing the White Limiter down about 20 levels brought up more H sharpening. I also found that making the radius value (Frequency) very high (95) also minimized halos to near invisible. But I am rethinking the whole strategy of sharpening in-camera and, pending results of an upcoming series of tests with outdoor footage, may abandon in-cam sharpening.

At the moment, my Blu-ray render of a parade that I shot with -3dB gain is the best image I've ever seen on Blu-ray. 90% of the major commercial releases I have in my Blu-ray collection are softer, grainier and many even have block artifacts, noticeable when paused, and MUCH subdued color saturation and certainly inaccurate color, often shifted to cyan. Surprisingly the parade footage looks good despite the frame being busy with lots of tree foliage and hand-held camera movements, not to mention people passing by just 36" from my lens. I'd imagine a cityscape or a sterile scene that's part of a controlled shoot would be a lot less burdensome on the CODECs. I'm continually amazed at how intense the color saturation can be (ie., bright red cars, fire trucks, shirts/uniforms) without crushing or losing detail. I'd reckon this would look awful over a composite connection, due to the sheer amount of primary color saturation, but over HDMI, it looks amazing. HiSat matrix with +20 Level is almost overpowering. I think I'll try STD matrix for a while and see how that looks--I can always play with color in post.

I've re-burned the BD-RE test disc about 6 times this week so far, and working in Encore has become much more predictable and friendly. Rendering this h.264 is proving to be time-consuming though, to say nothing of all the space uncompressed HD footage takes up on the array!
 
Thanks for your efforts Mark!

I have been buring Blu-rays for some events and have found the AVC/h.264 codec to look better than the Mpeg-2.

I use Edius to encode and Encore CS4 to author.

I also tried encoding in Encore but found the gamma level to be too high which results in a milky grey overtone on the finished image played on my plasma.

When I get some more time in a month or so I will dive in a bit deeper with all of this stuff as it is quite deep as your posts have shown.

These cameras and tools are amazing and with the right tweaking produce much better results than "out of the box" settings.

No doubt I will go back and re-read all of your posts on audio and video settings to get up to speed.

About in-camera sharpening, I dial it back on every camera I own. I learned this from using Panasonic cameras a lot and the main "detail" setting is really edge enhancement.

Turning down the in-camera stuff alows your much more advanced computer software have a shot at a softer edge image which looks more filmic to me.

The Mpeg-2 codec will do this as well with the edges as I found with my recent Blu-ray testing. The AVC/h.264 is a lot better with regards to edge fringing imho.
 
Last edited:
I have made some trhee camera shoots into a Blu-ray and noticed this with all three cameras, HPX-500, EX-1 & HMC-150.

Maybe it is because the AVC encoding allows more footage to be put on a disc thus allowing a higher quality level compared to the Mpeg-2.

The max for AVC for Blu-ray is around 20,000 kbps where Mpeg-2 maxes out at around 46,000 kbps for Blu-ray.

The discs I compared were at 18,000 kbps AVC vs 28,000 kbps Mpeg-2.

Still experimenting and I could be off base...
 
Barry J. Anwender said:
Detail OFF is prudent.

And do you see problems with your edges on the Mpeg footage sourced from the EX camera?
Click to expand...


I'll interject that I saw some artifacts at 200% magnification in Premiere, on a complex scene with foliage and water rippling.

Haven't gotten time to do it yet, but I'm planning to set up a test where I use various sharpening levels in camera, to detail off, and then try to sharpen up the detail off images in Premiere and see how the artifacting looks in both pre-and post-sharpened footage.

One concern I have about post-sharpening is that I have no tools to prevent halos from sharpening edges in post. I can do it in camera with the black and white limiters and crispening controls, but in Premiere I only have unsharp mask that's any real use and it sharpens everything indiscriminately. I could use a filter that sharpens only the midtones and leaves the highlighted edges alone. In this respect, the EX1's controls are superior.
 
True. An advanced sharpening tool would be a welcomed asset to all NLE's.

I use Edius and the sharpening is only varying amounts, but it is quite good. I increase it just to the point under where halos are formed in different contrast situations.

But a targeted sharpening tool would be better.

I think I will fill out a feature request for this.
 
I recorded some footage of some downtown scenes yesterday before fireworks and I used 3 setups: my usual +20 sharpening level, Detail OFF and +99 sharpening (just for stress testing the CODEC).

I opened the footage in Premiere CS3 and set the monitor to 200% magnification.
On the 99% sharpened clip, there were copious artifacts, both fringing and block (quantize errors) in the CODEC.
The moderately sharpened footage looked okay, but the Detail Off footage looked the most free from blocks and other nasties.

Next step was to apply some post sharpening to both the +20 clip and the detail off clip. I applied Unsharp mask in different amounts to each clip, with the goal being to achieve equal detail and sharpness in BOTH clips and then to observe artifacting.

I added significantly more sharpening to the detail off clip, obviously, to achieve same sharpness as in-camera sharpened footage. Both clips were super-sharp.

On the surface, at 100% view, it was apparent that halos were present around edges, like the trees. No halos on the post-sharpened-only footage.

There were some other subtle differences too: The in-cam sharpened footage seemed to lose texture detail in the foliage 1/4 mile away. A barber shop sign on the far side of the railroad tracks was a little more muddled. The post-sharpened only clip had more detail in the far trees and the barber sign was more clearly readable. I also noticed the tire treads on an SUV parked on the next parking lot were more clearly defined in the post-sharpened shot.

Another interesting thing was aliasing: the in-cam sharpened footage has an irregular pattern to the clapboard shingles on the far buildings where it alternates like thick and thin lines in a 'buzzy' and distracting manner, whereas the post-sharpened video shows the clapboard with consistent, even thickness on every row of shingles.

I think there was slightly more noise in the post-sharpened footage due to lack of crispening, but it's no issue compared to what is gained in overall quality.

Fringing seems not to be a problem with post sharpening and detail off. I am liking the look of what I saw, and in motion, rendered out one after the other to h.264, the in-cam sharpened footage has a slightly more 'watery' look to it, like the noise in the MPEG compression artifacts, whereas the non-sharpened footage was more solid looking and has less temporal 'noise' in it.

Yes, it would be great if we had more advanced sharpening tools in NLE, but from what I saw here last night, I'm not hurting by that!
 
Sounds like the theory of getting the cleanest image from the camera before it hits the codec is holding true.

It is quite amazing to see how much detail is in the image after putting on post sharpening. When the clips come in they look a bit scary then the magic wand makes them nice again.

Looking forward to some more testing with this camera as I know there is more image quality I can pull out of it if I put the time in.

I am very please with the controls in the menu. I have been cutting an HPX-500, HMC-150 & EX-1 for some shows I shot and I really don't need much color correction.

This is after I matched the EX-1 to the Panny look with my production monitor and a MacBeth chart.
 
It's amazing, but with tweaking, the pictures I managed to get on to Blu-ray look better than 90% of the Blu-ray movies I own, and at least as good as the few movies I have that are shot with Arriflex 435/235, Arri and Angenieux Lenses. Hyper-detailed, 20/20 vision-like clarity, and, unlike the movies, bold, brilliant colors and typically no visible grain.
 
I've been creating Blu-ray for years now, with various cameras including my current EX1. I've made burns on all kinds of media. Some of the tricks that exist include Blu-ray playback from USB flash drives with menu functionality and native 1080/24p playback, (something the PS3 won't normally do from other media besides optical disk, without some small enabling hacks.)

My perception is that optical media is becoming less relevant. It's possible to make AVCHD disks that play at reduced bitrates on most of the players now. Or you can use BD-R/RE with full bitrates. But to me, optical is a time waster.

I like authoring to the USB flash drives. It's fast, and you're not limited to redbook bitrates.

With the EX1, it's possible to get the video stream smart rendered onto a Blu-ray disk without ever rendering it. That's the ultimate quality from the EX because it's native. So that's where I also am in agreement with Basspig about getting the detail and other processing done in-cam, (correctly).
 
Interesting but a lot of consumers don't have an opitcal Blu-ray player much less a player that can play a USB device.

This should have been standard on all Blu-ray players but is not.

Who is going to buy what they can not play?

Also, the price of USB drives as a deliverable seem too high as I am using $4 BD discs.

Unless you can factor $99 into the price and give them a WDTV, I don't see how USB is effective for getting your content on people's televisions.

For computers it is the winner.
 
Bassman2003 said:
Interesting but a lot of consumers don't have an opitcal Blu-ray player much less a player that can play a USB device.

This should have been standard on all Blu-ray players but is not.

Who is going to buy what they can not play?

Also, the price of USB drives as a deliverable seem too high as I am using $4 BD discs.

Unless you can factor $99 into the price and give them a WDTV, I don't see how USB is effective for getting your content on people's televisions.

For computers it is the winner.
Click to expand...

Totally agree on all points, however you can't make money very fast burning your own BD copies, so the reason we are doing it likely is to say we can while previewing our own collaborations that few will see. It's the reason I can't get excited by it after doing it a long time now.

The exception is client/corporate/wedding video where only a few copies need be produced, but the end goal is still HDTV distribution. Whether that takes the form of Blu-ray, or a WDTV for a trade show booth, the video is still produced and targeted to the playback method available. The only mass distribution outlet that seems to be truly gaining much traction for HDTV content remains cable/satellite/ota.

If I were to speculate, that the video format that unseats Blu-ray may not even be another HD format, but perhaps a pair of 3D sunglasses marketed to a generation that grew up on ipod.
 
Bassman2003 said:
I have made some trhee camera shoots into a Blu-ray and noticed this with all three cameras, HPX-500, EX-1 & HMC-150.

Maybe it is because the AVC encoding allows more footage to be put on a disc thus allowing a higher quality level compared to the Mpeg-2.

The max for AVC for Blu-ray is around 20,000 kbps where Mpeg-2 maxes out at around 46,000 kbps for Blu-ray.

The discs I compared were at 18,000 kbps AVC vs 28,000 kbps Mpeg-2.

Still experimenting and I could be off base...
Click to expand...

I got a bit off topic in the last post, but on to your point above,

- I have found that AVC h.264 18 mbps vbr will playback on many Blu-ray players (as AVCHD) even from redbook dvd media, and that the encoded peak bit rates can reach 35-38 mbps without problem, but mpeg-2 from redbook dvd media will stutter at much lower peak bitrates on players with readrates of 1x. A few players like the PS3 that can spin the disk 2x can handle higher read rates for mpeg-2 as well. The bottleneck is the read speed of the optical drive. I don't really see why there should be a bitrate limitation of 20 mbps for AVC h.264 on BD-R/RE, although on the other hand, with EX1/EX3 there is little reason I've seen for why the AVC bitrate needs to be higher than that.

- The bitrate limitations should go away with BD-R/RE media, or when playback is from USB drives.
 
Last edited:
You must log in or register to reply here.
Back
Top