Photo Blog #8 - Bonus Torino Edition

I spent the weekend with my wife in Torino, Italy. Here are some images from there.

The transfer point between Zürich and Torino. The Milano train station.A galleria in TorinoClick below for more photos

The Belltower in TorinoWhat appears to be an old theater.Looking down a busy street

Another galleriaThe Mole Antonelliana.Down Via Garibaldi

Tips for Better Youtube Quality

A while back I posted an article to my old blog called '6 Tips for High Quality Youtube Videos.' Well that site doesn't exist anymore and Youtube is quite a bit different now so its probably time I update that article.

When posting to Youtube the first thing to remember is that it will make your video look worse. No matter what. Youtube is compressing your video and compression makes things worse. Period. However, if you follow some simple rules, you can minimize this worsening, or 'Youtubeification', to the point where it is not noticable. Read on.

1. Shoot the highest resolution and the lowest compression possible

There are about a million different video cameras on the market. There about 500,000 different video formats that they shoot in. Ok, I exaggerate, but your choices in shooting formats are immense. Some of these formats are better than others. To achieve the best results on youtube, choose the highest resolution and the lowest compression possible. For resolution, you should pick an HD camera (1920x1080 or 1280x720 pixels) over an SD camera (720x480 or 720x576 pixels). As far as compression goes, you want a camera that has the biggest file size possible once it gets onto your computer. Also, if you can see compression artifacts in your original video, these artifacts will get worse on Youtube. Ok, so you probably already own a camera and maybe its not perfect for Youtube. What else can you do to get better quality?

2. Use a Tripod!

That's Right! Using a tripod makes your video look more professional and it helps minimize the compression problems associated with Youtube. Why is this? Video compression is based on motion -- the more motion in the frame, the more work the compression has to do. When you shoot hand held, every frame is a little bit different which makes a compression algorithms go crazy. When you use a tripod, the image only changes when something in the frame moves which is a relatively small change. Or it changes when you pan or tilt, but these changes are smooth and predictable enough to be easy on compression.

3. Have plenty of light

Compression is based on detail. If things are dark it will be harder to achieve nice compression. This can be as simple as opening the blinds, turing on a lamp or going outside. In general, bright is better than dark for compression.


The following has changed due to Youtube's introduction of widescreen video. A new post is coming soon.

4. The secret ingredients to encoding for Youtube: at least 6000kbps h.264, 640x480 (640x360 for widescreen), Multipass encode, deinterlace.

No matter what you use to encode your video, these magical settings should insure a clean version to post to youtube. I personally use Quicktime Pro. So I will go over these particular settings here:

 

 

The Quicktime Pro Export Menu

Selecting the h.264 codec and setting

 

  • h.264 is the most modern codec available and is able to compress video in the best looking way possible.
  • At a data rate of at least 6000kbps the compression should be virtually invisible for a frame size of 640x480.
  • The frame size of 640x480 or 640x360(for wide screen) are chosen because this is the native frame size that Youtube displays
  • A Multipass encode checks each frame at least twice to ensure that it is encoded the best.
  • If you shoot on an interlaced format, you must deinterlace because computers are a progressive format. If you don't, you will see the so called 'mice-teeth' effect.
  • If you shoot on a widescreen format(16x9) then choose the output size of 640x360 (under custom in QTPro size menu)
  • One more tip: Try to get as close to the 1 gigabyte files size limit as possible by increasing the data rate. Don't go over board though. A 10 second clip won't ever be that big

There. Now you know the best settings to get the best out of Youtube. As a public service I have posted below two examples. One with the right way of doing things and one with the exact opposite. Enjoy!

This clip was shot with a Sony Z1U. It is a clip from my upcoming documentary "Legacy of the Great Aletsch"

This clip was shot with a janky $99 DV-DA1 VP crap camera. It is a clip of my cats shot a few hours ago.

John McCain Doesn't Understand Modern Media

He doesn't learn from his mistakes either.

Back in June, McCain gave a speech in Lousiana, his first after Obama had clenched the Democratic nomination, and arguably McCain's most important campaign speech to date. Well, the geniuses in McCain's campaign set up this nice stage for him:

Not only is green a bad color to contrast with a human face (it makes you look old and sickly -- umm...), this background also initiated Project Make McCain More Exciting wherein Stephen Colbert asked Americans to do just that. Watch:

Well, last Thursday, McCain gave his most important speech in his candidacy so far -- his acceptance speech at the Republican National Convention... and what do you know:
Come on! This time it's an even better green screen! To be fair, it wasn't like this the whole time. The background was a screen with revolving stock imagery, and this was the grass in front of a building. However, you would think that his team would have the foresight to know what the close-up camera angle was going to look like. Also, at points in his speech, the background was completely blue -- a background that is suitable for color keying as well. It wasn't too long before McCain was served up the political equivalent of the Rick Roll: The Barack Roll

These are the most recent and blatant examples of McCain not understanding how modern media works.  John McCain doesn't know how to use a computer. He doesn't understand that if you say something, then later say the exact opposite, someone will cut these statements together. Also, he doesn't realize that if you check out your new hot running mate's ass while she gives her introductory speech, everyone will see you. While on the subject of Sarah Palin: If you try to sell her as being a rebel against government earmarks, make sure she is not Queen of Earmarks. Back to McCain, he may not know how many houses he owns, but he could have checked Google Earth. The list of McCain obliviousness goes on, and this is just media incompetency. Imagine all the other ways he would be an incompetent president.
Now, I am not saying Barack Obama understands modern media any more than McCain (although he probably does). However, it is obvious that he can at least hire people who know not to set up a green screen behind him. If McCain can't even hire a decent AV department, what do you think his cabinet will look like?

Viva BasVegas

Tomorrow I will be filming a short promotional piece for Zürich fashion designer Basman during their fashion show entitled BasVegas. It should be a pretty cool event. If there is WiFi at the venue I will stream some of the show live on my Qik Channel. I will definitely post video and pictures later as well.

Film is Dead.

 

Well, maybe not dead -- but the revolution definitely will not be shot on film. I had the pleasure to shoot with a RED ONE for the first time this weekend and it really is the game changer that everyone says it is. I have been a fan of shooting video for a long time (I guess I am just of the digital generation), but the RED ONE  is the first digital video camera to really compete with film in all categories. For the record, I would like to make comparisons between Common video formats(MiniDV, HDV, Digibeta, etc), Film and the RED ONE.

Resolution

  • Video: Ranges from 640x480 pixels to 1920x1080 pixels with a variety of different pixel apsect ratios (rectangular pixels) thrown in there for good measure.
  • Film: No pixels, but comparable to 4k. Though most digital intermediates are done in 2k resolution
  • RED: 4k -- 4096 x 2304.

 

Recording Medium

  • Video: Lots of tape formats, some proprietary solid state and hard disk formats.
  • Film: 35mm or 16mm Negative Film.
  • RED: High Speed Compact Flash, Raided SATA drives, anything that accepts HD-SDI

 

Dynamic Range

  • Video: Varies wildly but averages between 4 and 6 f-stops
  • Film: Depends on the stock, but reliably 11 stops.
  • RED: Technically 11, but in reality around 9.

 

Image Plane Size

  • Video: Anywhere from Tiny up to 2/3rds inch.
  • Film: Super 35mm -- 24.89 mm × 18.66 mm
  • RED: 24.4mm x 13.7mm

 

Hardware Cost

  • Video: Starts from as low as a few hundred, tops out around 100,000
  • Film:  A new Arri 435 is around 75,000 body only. Prices vary in the + 20,000 range
  • RED: 17,500 body only.

Shooting Cost

  • Video: Some tapes and hard drives for editing. Quite cheap.
  • Film: Raw film stock, Film Processing, Telecine. Hard drives for editing. Quite expensive.
  • RED: Some CF Cards, Maybe a RED RAID, Lots of hard drive space -- probably a raid for storage. Not cheap, but cheaper than film.

 

Frame Rates

  • Video: Most commonly 24, 25 and 29.97. Some newer cameras can shoot up to 60fps
  • Film: As fast as your camera will let you go. Ultra high speed cameras can shoot 10,000 fps. Up to 120 is quite common
  • RED: Varies depending on resolution. Up to 30fps at 4k, 60fps at 3k, 120fps at 2k

 

In all these categories the RED is fairly equivalent to film, but beats film in terms of cost. A few other areas where I think the red wins are speed of use and safety of footage.

Ease of use: We shot a short film in 48 hours in 4k resolution this weekend. This would have been incredibly difficult with film. Film needs to be processed and telecined before you can begin editing. Both processes that would have taken valuable time. It probably would have been impossible here in Switzerland due to the lack of 24 hour facilities.

Safety of footage: Film advocates will say that a hard drive can crash or a CF card could fail. However, film is obviously more perilous to use than digital. Just one little light leak and your footage is ruined -- not to mention what can happen in the lab to make your work unusable. With file based digital shooting you can make as many back ups as you want on set. With film your camera original is all you have until you have a work print made.

One of the areas where I found RED more difficult to use than Film was focusing. The RED shoots 4k, but as of now, all of the monitoring options only go as high as 720p. This makes it difficult to see what is in focus. There is a pretty nifty focus assist that outlines in focus areas with red lines, but this is not perfect. That said, with a little practice, I was able to pull focus fairly well.

There are a few other great things about the RED: REDCODE RAW, Speed Ramping, Stop Motion, User definable buttons, Modularity, etc. Too much for me to go into right now.

I am ready to shoot on RED again as soon as possible. Last weekend makes my dread shooting HDV again.

 

PS. I know I left out cameras like the F23, the Panavision Genesis and the like. They have lots of the same benefits of the RED but so far none of them shoot 4K.

48 hour film competition

The talentscreen.ch 48 hour film competition is in its final stages. Just over 6 more hours remain until we have to turn in our film. I am failry happy with how it turned out. Shooting with a RED ONE from the very generous rental company Octamas was the best thing about the shoot. It makes me never want to shoot HDV again.
The competition began at 12:00 on Friday. All of the teams met at the UTO Kino in Zürich and were given the theme for the competition. This years theme was "Dreck am stecken," which roughly translates to "Dirty Hands" implying some kind of guilt. After we got our theme we spent about 3 hours working out the story and another 4 hours or so organizing the location, actors, props, wardrobe, etc. We arrived at our location at around 8:00pm and we shot until 8:00am.
With all the shooting done we decided it was best that we get some sleep, and start editing at 5:00pm. We have been editing ever since. I would say that we are about 97% completed right now.

Stay tuned, I will post some frame grabs later.


PAL HDV to NTSC DVD

Well my Hell is over.I have been trying to figure out a sure-fire method of creating a NTSC DVD from a 1080i50 HDV source. I think I have finally discovered the best method. If you are trying to do this and its driving you crazy, then try following these steps. (this requires Final Cut Studio 2)

1. Export your final movie as a 1440x1080 50i Prores 422 file. Do this by going to file->export->quicktime movie and then selecting the Apple Prores 1440x1080 50i setting.

2. Open this quicktime in Cinema Tools. Select conform->23.98. This changes the frame rate by making the video playback 4% slower. This will be perceivable in the audio and will be fixed later.

3. Drag the conformed quicktime to into Compressor. Drag the Apple ProRes preset onto the file. Change the settings to:

  • In the frame controls pane select
    - Resize Filter: Better
    - Output fields: Progressive
    - Deinterlace: Better
    - Everything else at its defaults
  • In the geometry pane
    - Dimensions: Select 720x480 from the drop down
    - Pixel Apsect: Select NTSC CCIR 601/DV (Anamorphic)
4. Submit your compression job. For me this took approximately 4x realtime.

5. You now have a NTSC 24p ProRes 422 file. Import this into final cut pro, make a new sequence and drag it in. When it asks if you want to match the sequence to the file say yes.

6. Double click the audio in the sequence and apply the audio filter AUPITCH. Change the first setting called 'pitch' to 80. Leave everything else alone.

7. Export this file as a quicktime movie with the default settings

8. Bring this file into compressor. Drag the default settings for DVD encoding onto your movie (either 90min, 120min or 150min Best -- pick the one that your movie is less than or equal to)

9. The automatic settings should work but double check that it is making a 16x9 progressive NTSC movie. Submit you job.

10. Congratulations. When you are done this should make a nice NTSC DVD. It is not that commonly known that DVD NTSC can be 24p but in fact most commercial DVDs are 24p. All dvd players are able to convert the video to 29.97 in real time.

A couple of notes:
This work flow was for a project shot on a Sony Z1 your camera may vary.
If you have extra time you can de-interlace with the "best" method. But this takes MUCH longer and I think the benfits are negligible.
If you have ALOT of extra time you can do this all in one step by importing your HDV into compressor and changing the framerate to 29.97 with Framerate conversion set to 'Best.' But this take A VERY LONG time. My 55 minute project was going to take over a week on a top of the line Mac Pro. I have used this method with shorter projects at the results with this new method are the same if not better.

If you have any questions feel free to leave a comment. I a relieved to have figured this out and am happy to have shared it.

   

DVD Creation Hell

Pal? NTSC? HDV? Prores422? 23.98p? 60i? 50i? This is what I am dealing with right now.

What I am trying to accomplish sounds quite simple. I want to make a 16x9 NTSC DVD from the PAL HDV master QuickTime for my new documentary Legacy of the Great Aletsch. However this is not as simple as it sounds. There are several thing to consider.

First is the frame rate. I need to convert the original frame rate of 50i (50 interlaced frames per second) to either 60i or 24p. However both options have their problems. Increasing the framerate means that we need to create new frames out of no where. This is not a simple process and often introduces stutter artifacts. Lowering the frame rate means we lose some frames here and there. This can introduce a different kind of stutter.

The next thing I need to consider is frame size. HDV has a framesize of 1920x1080. NTSC 16x9 has a frame size of 720x404 Actually, even this is not so simple because both of these formats use non-square pixels. So HDV's real dimensions are 1440x1080 with wide rectangle pixels and NTSC is 720x480 with squished pixels. So I have to go from one non-square pixel format to another.

The final problem is interlacing. HDV is an interlaced format, meaning that each frame is seperated into two fields made up of the odd and even lines of the frame. With NTSC we have two choices. I have to choose interlaced or progressive.

So I have a multitude of choices to make. Do I convert from 1080i to 480p or 480i? Do I change the framerate first and then downconvert or the other way around? I also have to decide which piece of software to use during the conversions and which codecs to use. It is maddening to figure out which gives the best results. I will update when I have found the best workflow.