Monday, October 26, 2015


It took a lot longer than expected - that's because I'm trying to learn video editing software as I go. It's also filled with mistakes, but I'm leaving it as is and calling it a learning experience. There are parts of it I'm happy with.. 

Saturday, October 24, 2015

Filmmaking System part 2 -- concerning the remote monitor

New stuff I've learned and need to remember:

  • Don't hot-plug HDMI cables! Have both devices (camera and monitor) switched OFF when plugging or unplugging cable. Hot-plugging can damage the electronics, and will most likely cause the camera and monitor to both shut down suddenly after working for a little while. 
  • Don't change anything on the monitor after recording begins or the monitor will shut down instantly and the camera will stop recording.
  • Always remove battery from back of monitor when it's not being used - it drains even while shut off. See about getting a power adapter for it since it gets used only in the studio. 
  • If you forget and hot-plug the HDMI cable or change a setting on the monitor and it screws everything up, you probably need to re-format the memory card before camera will record properly again.

Sunday, October 18, 2015

Putting a system in place for filmmaking

Finished the solenoid actuator bracket late last night after a marathon two-day build - got hours of video to edit down to a montage that will go live today or tomorrow. And learned lots of lessons about how to do filmmaking right (and wrong).

Notes I jotted down during the shoot:
  • Never use the GF3 when you can use the GH2
  • Go to the extra trouble to do each shot right - don't get lazy and end up capturing crap shots.
  • When going in for a closer shot change to a different viewing angle as well.
  • Move camera around a lot - you don't want an entire film shot from a single viewing angle.
  • Be creative with camera angle and lighting.
  • Use B camera to capture incidental stuff to fill in the story.
  • Shut off music while filming - otherwise you'll need to replace the entire sound track. 
  • Don't sing or whistle or say dumb stuff while camera is rolling - see previous item.
  • Clean up areas that don't look good. Clean them thoroughly - dust, cobwebs and dog hair will show up clearly in HD.
  • Make SURE camera is properly focused!!! And make sure it's recording.
  • Look at the lens - don't look at yourself in the monitor! (when shooting yourself)
  • Pull battery before it's fully discharged and pop it in the charger - have 2 fully charged ones always ready to roll (or 1 charged and 1 charging). Also have spare memory cards close at hand. Need to come up with  system to temp mark batteries to show if they're charged or discharged at a glance. 
  • Always transport memory cards in protective cases.
  • Before hitting record button, always check time remaining and battery life.
  • Always double check that every shot from the card has been transferred to the computer, then clear the card, and always format in camera as soon as the card goes in. 
  • Don't sit and watch footage as it transfers - let it do it's thing and get back to work.
  • Have card already in computer before opening Final Cut or it will crash. Always save project after importing new footage.
  • Final Cut is pretty stable while rendering or exporting - you can hide it and use the computer for other things. But it's very unstable while ingesting footage - don't try to make it multi-task at that time. 
Man, shooting live action stuff is a huge hassle compared to doing stopmotion! Just the sheer volume of material you need to capture and whittle down - I filled up an external hard drive that had over 300 gigs of free space!! I've ordered a much bigger hard drive for this, and I definitely need to learn more about how to set up a Final Cut project and work with the files, delete what you don't need as you go etc. 

Thursday, October 15, 2015

Combining footage from different cameras/ different frame rates works

Tested several different combinations today - footage from the GF3 and GH2, both in 1080p mode, and both AVCHD, but one camera was shooting 24 fps and one was 30. Final Cut had no problem importing and working with the different frame rates (well, it actually crashed the first time when I tried to import the 2nd card, but after that it went smoothly). Quicktime plays it no problem. I opened the inspector to see what framerate it was using and it seems to have exported the whole thing at 24fps, which was the setting from the first piece of footage. So it seems it used that info to create the project and conformed everything to it. I even tried uploading it to youtube and it worked flawlessly.

I also tried combining AVCHD and Mjpeg footage, both shot on the GF3 and at the same size and frame rate. That also worked, though I didn't try uploading it to youtube.

Since last night I've been trying to figure out how I was getting 24p from the GF3 - can't replicate it today. Possibly I was just mistaken, or maybe it was in one of the photo modes (you can set it up in stills mode and press the dedicated video button and it will capture video using the stills settings). I'll keep messing with it and if I find it I'll write it down before I forget.

Anyway, now I know I can use a 2-camera setup to record things and the footage will combine flawlessly into a single video -- as long as it's all the same size (didn't even try combining different sizes).

One thing I've discovered is that I now need to come up with a much better system for logging my footage and storing it - I literally need to write down what I'm recording on each card and create a nested system of folders on a scratch disc to contain it all, as well as a systematic naming convention for it all. It's weird - adding just 1 camera to the production line suddenly multiplies the logistical difficulty by several factors.

Wednesday, October 14, 2015

Hacking the GF3 - Mjpeg 720/30p mode is now 1080/24p!!

Today I put Flowmotion v 2.02 on the GF3, which was formerly a really good stills camera (with a great intelligent auto mode) but a crappy video camera, with soft mushy details all over. Now it's a turbo-charged movie machine with crystal clear visuals, and the Mjpeg 720p mode is magically converted to 1080p! Not only that, but whereas it's normally 30fps, it's now 24. Perfect - it's a cinematic badass now, and with one of the 64 gig 95MB/s cards it will give an hour and 24 minutes (or was it 40-something?) of pristine video.

It's still no good for stopmotion - won't take a remote of any kind and can't be set up to full manual in any mode. Of course I could rig the solenoid actuator on top of it to push the shutter button physically - in fact that's the original reason I bought the actuator. But no manual controls is a deal breaker. So it's now my little fun camera, for running around and just playing. It was ridiculously cheap - $58!! I can't help but wonder if the seller made a mistake converting from Yen into Dollars and left off a  decimal place, and maybe when the money showed up in his account he was slapping himself on the forehead over and over or something. Or maybe he inherited his dad's camera and just wanted to clear out some attic space, I don't know. At any rate, it's mine now and so cheap that I have little fear of damaging it (except for the lens of course).

Spent a good part of the day familiarizing myself with some of the video features - man, today's digital cameras are freakin complicated!! So much to try to remember - I think I'll just keep it in intelligent auto and try to discover what it can do for now.

I must have been capturing video from one of the stills modes like aperture or shutter priority, because I just checked and in movie mode the Mjpeg 720, while it has switched to 1080p, is in 30 fps, not 24. Complicateder and complicateder! I suppose I need to do a bunch of tests and write down the results, make myself a cheat sheet so I know what's going on. But then, since this is really just a run & gun camera I'm not concerned over what framerate it's using, just that the footage looks good.

Tuesday, October 13, 2015

Testing for my latest project

I want to be able to take pictures with my foot..

Saturday, October 10, 2015

Actual visible benefits of hacking the GH2 and how to access them

Ok, I'm in deep research mode and the blog is once again my notebook. 4th info-heavy post today:

This is from a thread about the actual visible differences between the hacked and unhacked GH2. I need to understand it and learn how to take advantage of the benefits:

"The hack version brings some subtle for some but big improvement to someone like me. There are three aspect where I find big big improvement, I put them in how the hack unfolded.

The first thing was the codec in motion (not shown at all in the above test). The hack does improve the image quality when in motion, you will see it much better in high detail scene. No more image breaking up when panning etc. Some have developed low gop settings that they claim result is much more filmic (with the latest ptool there is a bug which is causing some inconsistencies, while it brought 1 gop... avcintra territory).

The second thing is in the shadow details ala dynamic range. Where the base codec was smoothing any underexposed area, the hack just preserve the detail below (it even preserve the noise!!!). Just boost your gamma in post a little and see how those macro-block just pop up. Now you can meter to a certain level for the highlight and recover by boosting the shadows in post, effectively the hack has increase the dynamic range of the gh2. By 1,2, ... stops!!! until further test I don't know, but it is quite substantial and something huge for me. I would like to know (test) what happens to the highlight also, if the same principal applies because the codec is geared towards what I would call normal viewing. That is it cuts details where normally mister consumer won't see, more so in motion, to achieve higher efficiencies.

The quantization/gradient/banding setting from the latest ptool. This is another huge improvement from the hack. In the previous iteration of the hack it was seen that even if you could achieve some high bitrate on detailed scene, on lower detail scene the birate would decrease a lot and could be the cause of the gradient/banding on large surface of flat colours . The prime example being blue sky where the gh2 exhibited a lot of banding, even a bit more compared to other 8bit camera. Again I think it is the consumerist nature of the avchd base codec, tuned primarily toward efficiency before quality. With the work of Chris and Vitaliy they have enhanced/change the base setting resulting in high and more constant bitrate across the board, and the gradient/ banding is greatly reduced.

When you take all of the above for me the GH23.VK is a much much better camera than the base gh2 (which was already good) and as someone at personalview who has a red says "I'm a RED owner and am truly amazed at what I'm seeing" . Now, if you are going to shoot some family or tourist shot the hack is surely unwarranted. But if as me you intent to do quite some image manipulation, heavy grading etc, the GH23.VK makes a huge difference. The base image breaks down much much faster than the hacked one."

Every little bit helps (bitrate info concerning the Sedna hack and stopmotion)

For my own edification, here's info on the hack I installed in my GH2. The important thing to note in each block is the bitrate - that's the amount of information captured by the sensor, which equates to greater detail resolution in the video.

Sedna A - Max Detail Q20 version, including the Pasadena Pulse audio hack

SDXC card - Sandisk 64Gig, Class 10, write speed 95MB/s (the bulletproof card)

Shot done in HBR mode (high bit rate):
FPS:                            29.97 (standard in HBR 1020 AVCHD)
Shutter Speed:            1/30 second  -   nice smooth panning - tested @ 1/25 and got jitter
Profile:                        Smooth, flat (everything set to - 2)
Bitrate on playback:   150.54 mb/s

Shot done in 24P mode:
FPS:                            24 (standard in 24p 1020 AVCHD)
Shutter Speed:            1/25 second   -  nice smooth panning
Profile:                        Smooth, flat (everything set to - 2)
Bitrate on playback:   160.97 mb/s
I had to laugh to see that the bitrate is actually higher in 24p mode than in HBR (high bit rate)! Those are really astonishing bit rates - pre-hack I was getting rates of 114.26 mb/s (it looked really good even then!) But here's the really crazy thing... I just checked the data on a couple of the clips shot on the old G1 in stopmotion, using still pictures assembled into a video file. Those stills of course are much bigger than 1920 x 1080. Here's the info:

Shot on the G1 in stopmotion:
Uncompressed full size video files:    Size: 2963 x 1240
Bitrate on playback:                             1,216.94 mb/s (!!!!)

Uncompressed HD video files:          Size: 1920 x 803
Bitrate on playback:                             376.72 mb/s (!!!) 

Wow!!! Even after reducing the size to HD, the stopmotion files are still more than double the bitrate of the Sedna files!! And this is on the G1 - the first Micro Four Thirds camera, @ only 12MP. It will be even better with the GH2, a 16MP camera.

shutter speed for 24p video - 1/25 of a second!!

And while I'm documenting things I learn so I don't forget them..

The standard procedure is to always shoot video with your shutter speed double the framerate. Ie - for 24 fps, shutter speed would be 1/50 of a second. Or that's as close as you can get anyway - on a PAL camera it would be exact because of the 25 fps framerate. But it actually works better to use a much lower shutter speed, which increases brightness considerably and gives more motion blur. And by lower I mean 1/25 of a second. Sounds crazy, and you'll get a lot of extremely pissed off shouting from people who learned the standard programming and have never tried anything different.

I first heard this on the Personal Views website, where people discuss hacking the GH2 and other Panasonic cameras - it's to combat the excessive strobing (judder, shudder or what have you) that you can get when panning rapidly using a high bitrate hacked camera - objects moving sideways tend to show up as double or even triple images and move in a jerky fashion. The standard way to deal with this in film and video in The Biz is to always pan very slowly - they even have a formula to figure it out (and you know how much I love formulas, right? Pfft!)

Because of the increased motion blur it actually creates a much more filmic look, at least I think. Though that might just be my initial impression - frequently when you first discover something really cool like this it delights you and then after you come down from that and see it with a more sober eye you realize it's gimmicky and crude - time will tell. Anyway, now I've got the info documented here (and this is exactly the kind of technical stuff that tends to evaporate from my brain quickly).

Megapixel lens ratings - 2MP = 1080 HD

Just went through a little learning curve and I want to document it for future reference. For the first time I ran across something called the megapixel rating for lenses. I didn't really know exactly what a megapixel is, but I do know that my G1 has 12 of them and the GH2 has 16. So just now I was on the verge of buying what looked like a really decent 6.5mm wide angle when I notice it's rated for only 2 MP. Wait, what? Seriously!!?? I mean, isn't that like - totally 10 years ago or something? Who has a 2 MP camera?

A little digging turns up that while many cameras have high MP ratings for still photos, 1080p HD is only just about 2 MP. Ok, actually it's 2.07, but apparently a 2MP lens is good enough for shooting HD video at full 1080p. Though I wonder if a hacked GH2 might be trying to capture more detail than a 2MP lens can resolve? Would using it basically negate the hack? Like putting a low flow shower head on a high pressure hose?

Possibly, to some extent. I don't know tech stuff (love that line from the Terminator - use it every chance I get). But then I have some 1/2" format CCTV lenses that resolve only a small image circle in the center of the G1's sensor, and need to be severely cropped, which reduces the resolution to probably around 1MP. I'm not a pixel peeper or a perfection junkie, I believe as long as it works then it's fine. Hell, I love a lot of old films shot on 16mm film handheld with bad exposure and all that stuff - it can give a film a visceral punch in the gut kind of feel as opposed to a careful, overproduced and over intellectualized feel. Screw it, I'm gonna get it.

No good for stopmotion - the really wide lenses capture stuff that's all around you and almost behind you - you'd need a wraparound set including a sky or ceiling and to have the camera jammed way back into it so you don't end up seeing the basement walls over the edges, and a set like that would make it hard to reach in and animate, unless of course you're building a massive set that the animator can walk around on like a Laika production. But anyway, it would be good for some live action video stuff.

"In a megapixel (MP) camera such as the GoPro Hero line, the focal plane spot size of the lens should be comparable or smaller than the pixel size on the sensor. So if you are capturing high MP media you should use a lens that will resolve the pixels for the resolution you are capturing. MegaPixel ratings of a lens relates directly to the image sharpness, the ability to resolve pixels at a particular resolution. Using a lower MP lens than the MP rating of the mode you're shooting in will still work, though the media may be a little fuzzy (less sharp).

For GoPro Hero 4 & 3 video modes, here are the pixel measurements:

4K = 3840x2160 pixels 8.3MP (16:9)
4K Cin = 4096x2160 pixels 8.9MP (17:9)
2.7K = 2716x1524 pixels 4.1MP (16:9)
2.7K Cin = 2716x1440 pixels 3.9MP (17:9)
1440p = 1920x1440 pixels 2.8MP (4:3)
1080p = 1920x1080 pixels 2.1MP (16:9)
960p = 1280x960 pixels 1.2MP (4:3)
720p = 1280x720 pixels 0.9MP (16:9)
WVGA = 848x480 pixels 0.4MP (16:9)

For GoPro Hero 4 & 3 photo modes, here are the pixel measurements:

12MP Wide: 4000x3000 pixels
11MP Wide: 3840x2880 pixels (11.1MP)
8MP Medium: 3200x2400 pixels (7.7MP)
7MP Wide: 3000x2250 pixels (6.8MP)
7MP Medium: 3000x2250 pixels (6.8MP)
5MP Wide: 2592x1944 pixels
5MP Medium: 2592x1944 pixels

To figure out the MP of a photo or video resolution, you would multiply the pixels by each other. For instance, shooting a video in 1080p means you'll use 2,073,600 effective pixels (1920 x 1080 = 2,073,600). That 2,073,600 pixels is 2.07MP, or "mega pixel". So that means that if you plan to mostly shoot in 1080p (like we do) you'll simply need a 2MP or higher lens. The higher the MP rating of a lens the sharper the captured media will be, assuming you're capturing at a high MP that you'll be able to notice any difference. Using a 12MP lens (such as the stock one) to capture 1080p (2MP) content is theoretically not that much better than using a 2MP lens. There are other factors which affect image quality, such as the diameter of the outer glass element, anti-reflective coatings and of course the quality of the optics used (i.e. glass vs plastic)."

Tuesday, October 06, 2015

Nice demo on camera movement

Explains what effect various camera moves can have on the story or the viewer's perception of what's happening. You need to know why you're moving the camera.

Sunday, October 04, 2015

Got a GH2 and learning editing - keyframing brightness filter

I haven't made any direct progress yet on the stopmo flick, still haven't got the power fixed. But I did buy a couple of cameras and start to learn Final Cut Pro. In the spirit of journaling - largely to preserve the info here in case I need it later, here's the quick rundown.

First I decided to get another camera. Not for doing the stopmotion - the G1 is plenty good enough for that, but to document the process, to shoot stills (with the G1 in shot) and video and maybe even some time lapse. So I needed something capable of HD video, and I thought it would be wise to get another micro 4/3s camera so I could use the lenses I already have. Checked eBay and immediately found a Panasonic Lumix GF3 going for the ridiculous price of $58!! - they're usually more like $400 to $600 body only, even used. So I snatched it up, got a couple spare batteries , and then made the disappointing discovery that it doesn't do time lapse and has no socket where you can plug in a remote shutter release. Well crap! In fact, a little more digging turns up that it was aimed at people used to a camera phone or point and shoot who wanted to step up to M4/3s. So made to be used mainly in full auto mode, though it does allow manual control.

I played with it for a while, and cooked up a crazy plan to devise some kind of electronic shutter release using a linear actuator, something I've seen the Mythbusters use a lot for pushing buttons. Researched that (ok, I begged on SMA and Thomas Nichol was kind enough to talk me through the whole thing) and bought the stuff I need to rig it together, including a beginner's soldering kit for electronics.

Meanwhile I had also bought a 7" monitor since the camera's rear screen doesn't flip out allowing you to see it when shooting selfies. And it turns out it will work with a monitor - sort of, that is until you start recording, at which point the signal shuts off and the monitor goes black.

That was the straw that broke the camel's back, and I ordered a Panasonic GH2 - a filmmaker's wet dream said to capture video as good as $5,000 cameras and nearly as good as the $10,000 ones. That's if you use a hacked GH2 - this Russian guy named Vitaliy Kiselev found a way to get into the coding and trick it out so people could start creating patches to upgrade various parameters of its performance.

The GH2s tend to go for around $600 - $800 body only, but I found one for $330. Started researching on how to hack it, downloaded all the stuff I need and when the camera came in I started getting familiar with all the menus and features, and suddenly realized it must already be hacked, because the ISO goes all the way up to 12800! That's definitely not factory. No idea which hack it has, but the amount of detail and clarity is astonishing. And it should get even better when my new SDXC card comes in - the write speed of the card determines the data rate while filming.

Anyway, aside from that, I've been recording some video every day and editing it into little movies - stupid ones at this point, littered with bad cuts and bad shots and every error imaginable, but this is how you learn. By the time I get to work on the stopmo flick I plan to have reached a point of competency with editing and getting the right kind of shots to allow for good editing and good storytelling.

Last night I was testing various lenses for low light performance while walking through the house, some dark rooms some lit, and outside a bit. And I figured out how to add a brightness filter and keyframe it to bring light up smoothly in the dark areas and back down in the light ones. Really sweet!