Tuesday, April 05, 2016

Lens curvature, and Embracing its Perspective-Distorting Goodness

When you think of GoPro video, you may often think of extreme sports, or maybe great life moments, but rarely narrative cinema. The super-wide "fish-eye" lens, which has everything in focus, goes against many learned instincts of the cinematic language of storytelling. In late 2009/early 2010, coinciding with the rise of GoPro and the HERO HD, the DSLR video revolution began. This enabled consumers, using full-frame sensors and wide open lenses, to achieve shallow depth of field video imagery at an affordable price. While DSLRs were said to achieve a "cinema look", democratizing film and television production, the GoPro revolution was all over YouTube and Instagram. This gets flipped on its head when the filmmakers behind the viral YouTube® hit Bad Motherf***er (somewhat NSFW) landed a feature film deal and three years later gets a wide theatrical release (April 8, 2016) for Hardcore Henry. Nearly 100% of the film uses GoPro footage, without resorting to a fish-eye "correction" to give a first person experience, with all the barrel-distortion goodness.
Viral video hit from Biting Elbows
STX Entertainment's first trailer for Hardcore Henry
What am I on about?  Barrel distortion or fish-eye, with horizons and buildings appearing curved and all geometry in the image is distorted, these are unwanted artifacts, aren't they?  While it's true that Stephen Spielberg and Janusz Kaminski selected vintage lenses for Bridge of Spies, resulting in a more curved look than all other contemporary features in 2015, but bent tank barrels and non-geometric windows were employed for deliberate artistic reasons, not technical limitations.
Images from Bridge of Spies
There are good technical reasons to embrace the curve. And for those wishing to dismiss a POV feature film as only a live action first-person shooter, you need to revisit video games, as they do not use fish-eye projections. Video games are rectilinear ("standard" lens with straight lines everywhere), just like 99% of the lenses you might use to shoot a wedding or a feature film, and video games typically render a field of view (FOV) narrower than a typical GoPro. There is a lot more going on within the technical achievement that is Hardcore Henry.

Disclaimers and context... I did not always see barrel distortion as good.  I had no idea if Hardcore (its original name) could be remotely viewable on the large screen when I met the filmmakers, led by writer-director Ilya Naishuller, even before the first draft of Henry was finished.  I was GoPro's technical contact for the film, it was my job to convince the team that the HERO3 Black (new at the time) was up to the task.  I was a little bit over my skis here, but maybe we all were (it was an exciting time.)

I was from the school that super-35 was good, but full frame sensors were better, and film making was a lot about revealing less through shallow depth of field and extreme letter-boxing. Yes, I'm an amateur film-making hack.  And the rectilinear look was part of this desired look for me. It was through my insistence that the Fish-eye Removal feature was added to GoPro Studio after hearing from some professionals that the "GoPro look" doesn't always cut well with other cameras, plus I personally wanted it badly -- as did many customers based on analytics data.  The hack in me said rectilinear was the only correct/cinematic look.  At the same time frame, I was also taking up the quadcopter hobby, and rectilinear "corrected" GoPro footage looked so right to me.

Cine Gear Fly-over from David Newman on Vimeo
(HERO3+ shooting with a Medium FOV, this is not as wide a GoPro can go.)

Yet something nagged at me, rectilinear was not perfect (whether a physical lens or a post-correction), as it looked weirder with wider the source lens, even with gimbal-stablized aerial footage.  In GoPro Studio we limited the correction to a centered 16x9 crop as the distortions become undesirable when the full horizontal field of view is preserved. Why was rectilinear considered the ideal lens, when rectilinear itself can look extremely distorted? Examples of rectilinear distortion below.

HERO4 Photo, Stock lens, with about 120 degrees horizontal FOV

Rectilinear conversion, preserving horizontal FOV

I prefer the HERO4 look, as it is more representative of reality, rather than the elongated image of the rectilinear correction, however both are usable.  To push rectilinear beyond its limits I have simulated an even wider lens using a spherical 6-camera GoPro rig (stitched in GoPro's Kolor AVP.)

Simulated Equidistant "Fish-eye" with 150 degrees Horizontal FOV

Simulated Rectilinear Lens with 150 degrees Horizontal FOV
I'm standing on the left edge of both images, they are from the same frame in the spherical source, I much prefer the way I look in the fish-eye image. A practical fish-eye or rectilinear lens would produce the same result. We can see that at some point a lens-curved image becomes easier to view, and the more natural looking image.   Determining that exact point becomes an interesting problem to solve.

Showing the distortion in a few stills doesn't showcase how bad the problem can be in motion, as is clearly going to be the case in a POV action film like Hardcore Henry. 
Surrounded by bad guys in Hardcore Henry
This pan showing that "you" are surrounded by bad guys would be much more disorienting using rectilinear (at the same FOV.)  Familiar objects, like human faces, should not dramatically change size or shape as you pan. This rough render below, using spheres as an analog for human heads, demonstrates the distortion on the edges is clearly apparent when switching from a HERO4 lens to a rectilinear lens.

Same pan speed, same HFOV, switching between different lens curves.
I've discussed this with many, and I've heard this a good number of times, "when I see the world, lines are straight." Its as if linearity is the only distortion that matters, and that rectilinear lenses must be the correct simulation for our vision of the world. Yet our eyes aren't seeing the world as a projected image onto a flat screen.  A rectilinear lens is only a correct imaging model when the viewer FOV matches the shooting field of view, as many gamers are well aware. By viewing these images in a narrow formatted blog post, you aren't seeing them anywhere near their intended viewing FOV.  This mismatch from viewer FOV to camera FOV, is where the perceived image distortion comes from. 

From http://www.decarpentier.nl/ on "Game Graphics, Physics and AI"





Extra: fun reading on gamer's FOV theories


If you were to view the rectilinear image of the red car (from the examples above), from a distance half its displayed width, the distortion would go away (might be hard to do if viewing on a cell phone or iPad, even a large may screen my require reading glasses to pull focus that close.)  The same is true for a fish-eye lens, except using a curved screen with matching FOV.  We are used to rectilinear, but it is just as distorted when projected on a screen with the wrong viewer FOV.  When using a longer lens (telephoto), for display on wider FOV screens, the distortion appears as shallower depth of field, and a flattened image (much less so on smartphone.) A wide lens (28mm or less on DSLR) will enhance the depth when viewed on a TV, but it will look and behave like a standard lens on an IMAX screen. Many perceived lens attributes are due to the perspective distortion when the images are displayed with a different viewer FOV. When selecting a lens, we are often using perspective distortion, altering our images subtly to help tell our stories. 

So distortion is good, and is a key part of cinematography. We have been using it in all our photographic and film projects since its inception.  Super-wide style shooting can have a very large disparity between the captured FOV and presentation FOV (theater, TV, smartphone, etc.)  This type of shot benefits from a different type of distortion, a different lens curve, that helps map one FOV to a more screen compatible presentation format.  GoPro has been riding this wave since HERO HD. 

Distorting an image from a wide FOV to a narrower presentation has it advantages in that motion becomes more stable, particularly angle motion (camera shake and pans) is less pronounced with a small viewer FOV.  GoPro users have often commented how good the stabilizer is, when we don't have one, the wide curve lens inherently appears more stable. For a wide rectilinear lens the benefits peak around 90-95°, beyond that some barrel distortion is needed to extended the stabilization properties. The table below shows the size of edge movement for a horizontal pan at different capture field of views. The HERO4 is inherently more stable from 60° and above, with rectilinear motion quality reversing at 100° and beyond. Below 60° both lens types have similar motion characteristics.
Edge Motion as Lens HFOV increases, for a narrow viewer FOV

Given all the distortion difficulties of presenting super-wide images at standard viewing distances, it should become more obvious why a film like Hardcore Henry used GoPros. Put simply, the wider the fish-eye lens, the more world motion (particular angular motion like pans) slows down. When shooting human POV, with body-mounted cameras, cinematography rules for ideal panning speeds are difficult to control.  Wide fish-eye lenses are significantly more viewable, resulting in less motion sickness than many viewer experienced with similar shaky-cam sequences in films like The Blair Witch Project.  So with fish-eye, the wider the lens, the slower camera motion, with rectilinear, the wider the lens, the faster the camera motion (at the edges, where we are most sensitive to motion.) 

Although the advantage to shooting action with a fish-eye lens is not obvious to many reviewers.
  • From deadline.com "It’s not for everyone, and a lot of viewers might be made sick thanks to the jittery, fish eye lens through which everything is viewed."  
  • From joblo.com/movie-news/  "For what it's worth, I'd say the filmmakers did a pretty damn good job at keeping a lot of it coherent, although I'm not crazy about the fish-eye view for an entire movie."
While it is a new look for a movie, fish-eye is a technical requirement for making this of type first person POV possible without damping down the action.  I even experimented removing the fish-eye from the Hardcore trailer (it was crudely done, so I will not share here), and the result just like the animated-GIF above was unviewable in rectilinear.

Side bar: So why 120°, and not 150° or 180°?  There is nothing magical about this number, however it does somewhat match the human view FOV for both eyes, while we experience near 180° when you consider the peripheral vision of left and right eyes, the combined stereo image is significantly less.  In VR applications, many of the modern head-mounted displays range between 100-120° horizontally.  For a GoPro camera, 120° has more practical origins. Image sensors are rectangular, and we use a 4x3 sensor, so that we can achieve a wide and tall FOV.  The good region (still sharp) of a fish-eye lens would be between 150 and 170°, so the horizontal FOV can only be between 120-136°.  
Typical action camera lens and sensor setup.
To achieve significantly more than 130
° horizontally would require the stitching of a multi-lens system as required for 360/VR capture. So Hardcore Henry 2 (I have no idea, or I couldn't tell you if I did) could use VR rigs, but the simplicity of a single camera capture would likely win out.     
Upcoming GoPro Omni

Shooting POV? Shoot GoPro. That would seem the obvious message to come from a GoPro engineer, but that is not quite the message I'm going for.  Yes, there are clear advantages to a fish-eye style lens when shooting very wide images, particularly if there will be a lot of motion.  Planning content for YouTube, you already know the GoPro look is working great, leave it as is with no "lens correction" required or desired.   But for a theatrical release, you could potentially optimize the curve in post, to be slightly less for the theatrical version. 

GoPro camera's themselves have had more than one look.  
  • GoPro HERO HD was almost an ideal equidistant fish-eye. With 170° diagonal in the 4x3 modes.
  • HERO 2 pushed the fish-eye look further, slightly more curved in the center, with similar FOVs as HERO HD
  • HERO 3/3+/4 introduced a less curved lens, improving with each camera generation, closer to a stereographic projection.  Not striving for as much in the diagonal, with little over 120° in the horizontal.  It was designed to keep the action of high-speed motion looking real.
The filmmakers once commented that their music video was hard to watch on the big screen, so they needed to develop new techniques to make Hardcore Henry possible.  However, there was also a subtle benefit switching from HERO2 to HERO3, having a lens curvature more suited for the large screen. 

When determining the amount of lens distortion, screen size matters. Or more precisely, viewer FOV matters. If you are producing content for consumption on a smart phone the perceived distortion is different than when targeting a theater projection.  Wide FOV rectilinear looks even worse on a smartphone as the distortion is more pronounced. Something new to think about, as the viewer FOV is increasingly more varied, from IMAX to video played within a Facebook News Feed, we may need to optimize lens curve for one, or produce multiple versions in post. Just as theatrical and TV releases have different color correction passes, POV sequences can be optimized for the average viewer FOV. 

The average viewing FOVs for the large range of viewing environments:
  • Smartphone video portrait: 10-15°
  • Smartphone video landscape: 15-25°
  • HDTVs: 25-35°
  • Movie theaters: 30-60° (unless you sit in the front rows, this can be very high, don't sit in the front rows of Hardcore Henry, Sharlto Copley considers these rows "the splash zone".) 
  • IMAX average of 70° (from IMAX® 101 Theatre Geometry Extended)
Using the front wheel from the two car images from above, I have re-projected both images to observe the distortion at different viewer FOVs.
Where you sit matters


For such wide source, from a smartphone to movie theater, the HERO4 lens holds up well, but at the viewer FOV increases (like in IMAX) you can see the benefits for reducing lens distortion.

Telling long form immersive stories from a human POV is hard. To approximate the human vision we need a very wide lens, VR headset or sit front row center of an IMAX theater.  However, VR and IMAX can come with some unwanted side effects; as the viewer FOV gets larger we are more likely to experience motion, and potentially motion sickness. Large vistas or slow moving scenes in either IMAX or VR, activating our peripheral vision can be a fantastic experience, but the kinetic action and whip-pan nature of a film like Hardcore Henry works because it doesn't fill our FOV. It works on the large screen from normal seating, middle rows and further back, it will work on your home TV and mobile viewing devices. It works because there is a lens distortion optimally remapping the wide, human like, perspective to more typical viewing environments. I saw the U.S. Premiere of HH at the SXSW film festival in Austin, I was sitting towards the back with a FOV of around 30° and experienced only the slightest unwanted motion illusions during the first part of the movie where the action was slower.  Once the action picked up, all motion issues were gone, the experience is one-of-a-kind and I did not want it to end. 

April 14th '16  P.S.  I just re-watched the movie on a much larger screen, at a closer viewing distance, and I had no motion issues at all.  Seems that first person action is an experience that the audience will quickly adapt to. With the 15 or so GoPro engineers that joined me for this viewing, several did experience what I felt in my first viewing, but only in the early parts of the movie.  One co-worker did have to step out the theater, so it is not for everyone.  It was extra fun seeing the movie again, so many clues I missed the first time around.  

---
Good further reading on lens curvatures: 
About the various projections of the photographic objective lenses by Michel Thoby
In Praise of the Fisheye (in video games), by Daniel Rutter
Perspective Projection: The Wrong Imaging Model, by Margaret M. Flect
FOV and the "fisheye" effect - learn the difference, Randomeneh (on Reddit)

Saturday, March 08, 2014

Protune 2.0

There is new firmware out for your HERO3+ cameras.  I'm going to expand on some of the new HERO3+ Black Edition features related to Protune:

  • New advanced Protune™ controls for Color, ISO Limit, Sharpness and Exposure
  • Changes the Protune default settings for Color to "GoPro Color" and Sharpness to "High" 

Protune History

Protune was originally introduced in a firmware update to HERO2, and it was developed to solve the need of professionals already using GoPro cameras in feature film and TV projects.  Protune added a range of modes previously unavailable in any POV camera system: 24p, log encoding, H.264 long GOP at 35+Mb/s. See my first introduction of Protune at NAB 2012 -- this video was six months before its release.

Here is my October 2012 blog entry discussing the HERO2 Protune availability, just weeks before the HERO3 launch.

With the HERO3 Black Edition, Protune was included from day one. There were some small changes: 24p was now a standard video mode (no longer requiring Protune to be active), and white balance controls where now offered including CAMRAW.  Protune White balance controls for HERO3 and HERO3+ are:

  • AUTO - Same as in the standard modes
  • 3000K - Locked white balance for indoor warm lighting with an sRGB color space
  • 5500K - Locked white balance for sun conditions with an sRGB color space
  • 6500K - Locked white balance for daylight overcast conditions with an sRGB color space
  • CAMRAW - Lock white balance with sensor native color space.

CAMRAW is the only non-obvious addition, it doesn't attempt to saturate the image to the reduced color gamut, but standard, sRGB. Shooting in CAMRAW improves the ability to cut GoPro footage with larger cinema cameras, but it requires more post color correction -- except within GoPro Studio which handles the color matrix required automatically.

My shooting tip:  I always shoot Protune CAMRAW. The subtler color image is nice to start color correction upon, but CAMRAW is also lower noise. To saturate any image the difference in color channels is gained-up, blue channel noise is crossed into green and red channels, and vise-versa.  This happens in all cameras, and it happens in post saturation, but with CAMRAW it is under your control.


The New Protune


Protune within the new firmware on the HERO3+ Black Edition has changed again.  The original Protune was the "pro", "tune" that GoPro designed for all professionals, yet you don't have all the same needs. Now it is the mode for pros to tune their GoPro cameras.

Note: The new Protune defaults are very different to previous releases.

Protune is accessed within the tools menu by select CAPTURE SETTINGS:


Use the Mode button to scroll to PROTUNE and press the shutter.

While previously there was just on and off, with the number of new modes we added a reset to restoring Protune to the default configuration.  The default configuration turns all of Protune off except for the high bit-rate.  Protune is still the way to get the least compression / highest image quality.  If you're new to color correction, but want the least compression (for high action, high detail video), then turning Protune on is all you need to do.  Protune does required quality SD media, see GoPro SD Card Recommendations.
The next menu item exposed (with Protune set to On) is the white balance mode.   I'm showing my favorite CAMRAW mode here.
The white balance menu is unchanged from the last firmware with the default set to AUTO.

The first new menu sets the encoding curve.  It is described in terms of the video's appearance.  I also use the FLAT Protune log curve.
However the default for this mode is GOPRO COLOR, which is the standard high contrast, high saturation for the classic GoPro look. This is different from earlier Protune implementations which only had the FLAT log curve encoding of the video image.  GoPro Color was added to Protune to help with broadcast / news applications that rarely do significant color correction, but still want the best compression quality possible.

The next menu item is not completely obvious, as isn't common in other camera systems. While I know many pro users want full manual exposure, the nature of a camera with no mechanical iris on a super wide F2.8 lens, makes that tricky. ISO Limit is a step in the direction for manual control. A GoPro's exposure is controlled through shutter speed and sensor gain. ISO limit will restrict the sensor gain to the value selected or lower.
For night scenes you don't typically want to gain up the shadows to 6400 ISO (that is the default.)  If you want dark to be dark, consider 1600 or 400 ISO limit.  I haven't decided my favorite yet, other than not using 6400.  This is an improvement over the older Protune which didn't limit the sensor gain.  For those accessorizing their cameras with variable ND filters, you can use the ISO Limit 400 as a manual exposure mode of sorts. Through the ND, slowly stop the light down until the camera output begins to darken, you are now running ISO 400 with 360 degree shutter (hint: shoot 48p, then in post drop every other frame for 24p at 180 degrees -- also you will also have 2X slow motion ready when you need.)
I wouldn't know that icon meant sharpness either, but it is hard to visually depict sharpness with a off/on level monochrome display. Previous Protune modes had camera sharpening completely off, and whereas the new default and standard modes have the sharpness set to HIGH.
While I'm a fan of the old Protune with everything designed for post corrections, including adding sharpness, but I'm liking the medium mode as nice balance.  It doesn't often need additional sharpening (GoPro Studio will not default to adding any,) and it doesn't have any too obvious sharpening artifacts.
The mode I really needed: EV compensation.  I shoot a lot of events under stage lighting with my GoPros, bright lights on faces, so often blows out.  This problem is solved through EV compensation.

The EV compensation ranges from -2EV to +2EV stops of compensation, defaulting at zero.  I like to use -0.5 as my default, ready for anything, but I have used -1.0EV or -1.5EV for stage events.
Here is Protune 2.0 using the default settings, a 1:1 pixel crop from a 1920x1080 image.  No color correction applied.  Image is nice, but we can do more.


Here is CAMRAW with sharpening at Medium and EV at -0.5.  The color of the petals no longer clipping, under magic hour sunlight (EV -0.5 and CAMRAW helped.) 
Here is the same image as above with the color matrix applied (automatically) in GoPro Studio, some white balancing, contrast, sharpness added, output through the Protune preview LUT. The final image is more true the original rose coloring (this wide gamut magenta flower.)

















My Favorite Configuration for HERO3+BE

While I've listed some of my favorite settings above, here are defaults for all my HERO3+ Black Edition cameras (one is not enough):

  • Video 2.7Kp30 (24p for night events or creative projects) with a Medium FOV (this mode is the sharpest with highest resolution for video presentation, with very little lens curvature.)  For aerial video I use 1080p60 Medium FOV (low light off.)
  • Photo Burst 5/1 at 12MPixels
  • Timelapse -- I don't use time-lapse photo, I use 2.7Kp24 or 1440p24 video modes and compute my timelapses is post (a GoPro Studio feature.)  See Rethinking Time-lapse
  • Protune On
  • FLAT - Log curve
  • CAMRAW
  • ISO Limit 400 or 1600 as needed.
  • Sharpness Medium
  • EV -0.5 (or downward as needed)
  • Sound/Beeps OFF
  • Auto shutdown after 120 seconds of inactivity.

Added March 9th, 2014

GoPro App (Android and iOS)

These new Protune controls are all available through the latest GoPro App (free.)  An added bonus: within the latest GoPro App (version 2.3) will update your HERO3+ camera for you -- no need to update through the web site.

Hint: for updating the camera software via the App.  Get the latest app, and connect to the camera as usual.  Now the App knows which model you have, and it will contact the server to request any camera software updates.  This can take up to 24 hours of connecting to your camera (behavior from iOS and Android differs slightly.)

The next time you connect to your GoPro, this "Install Update" will appear -- click and follow the clear instructions.


Hint 2: For those that just can't wait 24 hours, after connecting to your camera, switch back to internet connected WiFi (i.e. not the camera.)  Then open the GoPro App and go to the App Settings (top right gear icon.)  Toggle the Auto Download on and off -- that should start the camera software download from the server.  See what will be updated under Camera Models.

Thursday, February 13, 2014

QuickTime 16-bit

This might be beating a dead horse, but QuickTime truly sucks.

For those using 16-bit (deep color) applications, always use the Force 16-bit encoding option, it is the highest quality and surprisingly, it is often the lowest data rate.

Now for the weird reason.

QuickTime loves 8-bit, it prefers it greatly, and support for deep color is difficult at best.  Over the years we tried to make the 16-bit the preferred mode for our codec within QuickTime, yet there are many video tools that broke when we did this.  The compromise was to add the Force 16-bit into the QuickTime compression option, to allow user to control the codecs pixel type preference – applications that can handle 16-bit will benefit, and applications that don’t, still work.

Using After Effects for my test environment (but the same applies to other QuickTime enabled deep color applications.) I created a smooth gradient 16-bit image, then encoded it at 8-bit using using a 8-bit composite, 16-bit using a 16-bit composite and 16-bit using a 16-bit composite with Force mode enabled (pictured above.)
Without post color correction, all three encodes looked pretty much the same*, yet the data rates are very different.

* Note: QuickTime screws up the gamma for the middle option, so with the image gamma corrected to compensate, they looked the same.

The resulting file sizes for 1080p 4:4:4 CineForm encodes at Filmscan quality:
8-bit – 13.4Mbytes/s
16-bit – 28.4Mbytes/s
16-bit Forced – 5.3Mbytes/s

Our instincts that higher bit-rate is higher quality will lead us astray in this case.

Under color correction you can see the difference, so I went extreme using this curve:
To output this (16-bit Forced)
The result are beautiful, really a great demo for wavelets.

Zooming in the results are still great. Nothing was lost with the smallest of the output files.

 Of course we know 8-bit will be bad
We also seeing the subtle wavelet compression ringing at the 8-bit contours enhanced by this extreme color correction.  This is normal, yet it shows you something about the CineForm codec, it always uses deep color precision.  8-bit looks better using more than 8-bits to store it.  That ringing mostly disappears using an 8-bit composite, an 8-bit DCT compressor could not do as well.
Storing 8-bit values into a 12-bit encoder, steps of 1,1,1,1,2,2,2,2 (in 8-bit gradients are clipped producing these flat spots) are encoded as 16,16,16,16,32,32,32,32, the larger step does take more bits to encode – all with the aim to deliver higher quality.  Most compression likes continuous tones and gradients, edges are harder. Here the 8-bit breaks the smooth gradients into contours which have edges. The clean 16-bit forced encode above is all gradients, no edges, result in a smaller, smooth, beautiful image.

 Now for the QuickTime craziness, 16-bit without forcing 16-bit.
The image is dithered.  This is the “magic” of QuickTime, I didn’t ask for dithering, I didn’t want dithering. Dithering is why the file is so big when compressed.  QuickTime is given a 16-bit format, to a codec that can do 16-bit, but sees it can also do 8-bit, so it dithers to 8-bit, screws up the gamma, then gives that to the encoder.  Now nearly every pixel has an edge, therefore a lot more information to encode.  CineForm still successfully encodes dithered images with good results, yet this is not want you expect.  If you wanted noise, you can add that as need, you don't want your video interface (QuickTime) to add noise for you.

If anyone can explain why Quicktime does this, I would love to not have users have to manually select “Force 16-bit encoding”.    


P.S. real world deep 10/12-bit sources pretty much always produce smaller files than 8-bit.  This was an extreme example to show way this is happening.


Sunday, December 30, 2012

Rethinking Time-lapse

Since I got my hands on the HERO3 Black Edition, I've been doing significantly more time-lapse.  This is not because the stills are so much better than HERO2, and they are, it is due to the new video modes at 2.7K and 4K, combined with Protune. There is no need to time-lapse with JPEG sequences with this camera, unless you know you need a very long interval.  Most time-lapse shoots that document human experiences, are better with shorter intervals between 0.5 and 10 seconds. Knowing what interval is best takes practice, but forget that! With the new camera shooting 4K at 12 or 15fps, basically a continuous 9MPixel motor drive sequence (stored in an MP), time-lapse guess work can be left to post. Resampling 4K video to your needed interval is a straightforward process in most video tools, and a standard feature in the free GoPro CineForm Studio software.

At 4K you get most of the spatial resolution of the still mode, at 6+ times the temporal resolution (12/15fps vs a maximum of 2fps for JPEG) at approximately half the data rate.  So there is a data rate saving using the video modes for simulating the shorter intervals, but for long intervals there are still reasons to shoot ultra-HD video over stills -- simulated longer shutter intervals.  The HERO cameras mainly use the shutter speed to control exposure, which is fine for high action moments, but for scenes best for time-lapse a fast shutter may not be desirable.  With a DSLR, you can stop down the camera's aperture, but that only gets you so far. For time-lapse exposures of 5 seconds of more, that would require a lot of neutral density filters for daylight shooting.  For a HERO camera, aperture control is not available, and adding neutral density is highly impractical, so we need do the camera operation in post.

So let's say you want to simulate a 5 second exposure with a 10 second interval in full daylight (simulating a 180 degree shutter at play speed.)  HERO's default exposure might be around 1/1000th of a second in strong daylight, nowhere near the 5 second exposure target.  Yet the camera could be recording 4K at 12fps over those 5 seconds, collecting 60 individual frames.  If you average those 60 frames, you get very close to the look of a single long exposure from a DSLR with a hell of a lot of ND filtration, without the setup headache. Typically blending over 30 frames for daylight simulates the motion blur of a single exposure.  With darker shots that might have the camera's shutter exposure near 360 degrees (1/30th for 30 fps video), far fewer frames can be blended for a natural look.  Of course, the more frames used in averaging, the smoother the results.  I have been asked how a GoPro achieved this high action shot with so much motion blur:

Now you know.  This was shot 1920x1440 at 24p, with 30 frames averaged for each single frame in the time-lapse output.

Continuing with the target of 5 second exposure and with 10 second interval, I was intending to model 180 degree shutter, however the CineForm Studio software with the Motion Blur enabled will simulate 360 degree (this was by design.)  So setting "SPEED UP" to 60, "FRAME BLEND (MOTION BLUR)" on and the output frame rate to 23.976p, the result will be a clip with 5 second exposure and a 5 second interval.
To get this to simulate a 10 second interval simply place it in your editing tool's timeline and double the playback speed (with frame blending off.)  Now every other 5 second exposure will be displayed for 180 degree shutter emulation.

Protune helps greatly, particularly in low light. Now that we are averaging frames together, we get an excellent side effect:  a large reduction in noise. Each doubling of the number of frames averaged will half the noise in the image. Combined with Protune, which preserves much more shadow detail, you can basically see new details that would normally be lost to noise with regular video or stills time-lapse.  Protune lifts the shadow detail so that it is no longer crushed to black. In standard mode, averaging crushed black only results in more crush black, yet in Protune averaging a noisy shadow detail results in more shadow detail.

I've used this technique in most of my recent videos, such as this one (the night time-lapses are very clean, because of HERO3 Black and this averaging technique using CineForm Studio):

24 Hours of Lemons at Chuckwalla Dec 2012. from David Newman on Vimeo.

Update Jan 5/2013: Example comparing classic and video blended time-lapse

P.S. For those who have been following my Instagram feed (http://instagram.com/0dan0) or via Twitter (@David_Newman), you are likely aware that I've been combining the above time-lapse technique with a motion controller I have been experimenting with.  This combined a GoPro with a 3D printed motion controller that runs on toy G-scale train track. I've just posted its design on thingiverse.com, for this project. Let me know if you successfully build one, and link me to your videos.

Tuesday, October 16, 2012

Why I shoot Protune -- Always!

If you are reading here and you don't yet know about Protune, read this entry first : Protune

Here are some images that should speak for themselves.  This wide dynamic range scene, outside lighting to the right (frosted glass windows,) indoor shadows in the back left, showcases the improvements the Protune curve offers for color correction:

Stock mode converted within CineForm Studio 
(Premium version of Studio added the waveform.)



Protune mode automatically corrected when converted in
CineForm Studio to be similar to stock
(check out those improve highlights.)



Stock mode with contrast reduced in Studio to show the
dynamic range limitations.









Protune mode with the same contrast applied as above 
(more shadow detail, reduced highlight clipping.)







Reseting Protune to no automatic corrections with
CineForm Studio, also gives you a nice starting 
place for color correction.

P.S. I shot this with a HERO3 Black Edition.






Wednesday, October 10, 2012

Protune

If you haven’t heard already, Protune™ firmware and GoPro App for the WiFi BacPac are now available -- get them now.  While my team worked substantially more on developing the GoPro App, this blog is about the origins and design of Protune.

Before I geek out on why and how Protune is so cool, some readers may want to know what it is and does (and may want to skip the rest.)  Protune is a suite of features designed to enhance an even more professional image capture from your GoPro, while still being accessible to every GoPro user.  Protune has the strongest emphasis on image quality by increasing the data-rate (decreasing compression) from an average of 15Mb/s to 35Mb/s.  Small artifacts that can occur in detailed scenes or extreme motion are gone at 35Mb/s.  Next is adding the 24p frame rate (to the existing frame rate options), greatly easing the combination of GoPro footage with other 24p cameras, common to professional markets. Finally the Protune image is designed for color correction; it will start with a flatter look that is more flexible for creative enhancement of the image in post-production. With the latest HERO2 firmware installed, Protune is enabled with the secondary tools menu.

Now for the why and how.

Protune has been a long time coming, and so has this blog entry.  Protune is an acknowledgement that so many GoPro cameras are used for professional content creation – Discovery Channel looks so much like a GoPro channel to me.  Protune is also the first clear influence the CineForm group has had on in camera features, for which we are super proud, yet most of the engineering was done by the super smart camera imaging team at GoPro HQ.  For the novice Protune user, CineForm Studio 1.3 is setup to handle Protune image development, so all users can benefit from this cool new shooting mode. This synergy between the software and camera groups, allows us to push both further. In the old CineForm days (non-GoPro) I would have probably blogged about helping with the design of a new camera log curve, and all the pluses and minus of color tuning, months before we would have had anything to show, but that was before we became part of a consumer electronics company.  Some things must remain secret. Working at CineForm was exciting, but it is nothing compared to the adventures I’ve already had at GoPro, with so much more to come. 

Protune for me started when HERO2 launched.  Here was a camera that I could use in so many ways, yet in certain higher dynamic range scenarios (I shoot a lot of live theatre and was experimenting with placing GoPros around the stage), the naturally punchy image limited the amount of footage I could intercut with other cameras.  It is of course the intercutting of multiple camera types that is of greatest need for the professional user. Note: there is one professional group I know of that exclusively uses GoPro HEROs, and that is our own media team – even though they now use Protune shooting modes. Protune gets you more dynamic range, and I was amazed how much. 

Sensor technology continues to grow, and we are seeing awesome wide dynamic range images coming from premium cameras like ARRI Alexa and even the amazingly affordable Blackmagic Cinema Camera, but as sensor size (really pixel size) shrinks, there is an impact on dynamic range.  Smaller pixels often result in reduced dynamic range, yet so much has changed in so few years.  Back in 2006, CineForm was very much involved with Silicon Imaging and the development of the SI-2K camera, which was highly praised and generally confirmed to have around 11 stops of dynamic range – good enough to be used on the first digitally acquired feature (well, mostly digital) to win Oscar Cinematography and Best Picture awards.  The HERO2 sensor is smaller and has significantly higher pixel count (11MPixel versus SI-2K’s 2MPixel, HERO2 pixels are way smaller), yet we are also seeing a similar dynamic range.  

It was not just five years of sensor technology that made all the difference, it was using a log curve instead of contrast added to Rec709 with 2.2 gamma -- geek speak for calibrating cameras to make the default image look good on your TV.  Making images look great out of the box is the right thing to do for all consumer cameras, and you get just that with HERO2 via HDMI to your TV. Yet TVs do not generally have 11 stops of dynamic range, maybe 9 on a good set, and that is after you’ve disabled all the crazy image “enhancements” TV defaults to having switched on (which typically reduce dynamic range further.)

So why shoot wider dynamic range for something that may only be seen on TV, computer monitor or smart phone (all decreasing in dynamic range)?  The question is somewhat obvious to professional users, as color correction is part of the workflow.  Color correction simply works better with more information from the source for which to choose the output range. Even the average consumer today is more open to color correction of an image thanks to the likes of Instagram filters. The more dynamic range you start with, the better such stylized looks can work.  Our own media team wasn’t using great tools like Red Giant’s Magic Bullet Looks until shooting Protune, which greatly increased the creative flexibility of the GoPro image output.

So why a log curve, rather than just reduced contrast with the regular gamma?  This is a trickier question.  The full dynamic range can be presented with a 2.2 gamma of standard TV, it will look a little bland (flatter or milkier) just as log curves do on a TV without color correction, so it holds no aesthetic advantage over log.  Log curves do have an advantage over gamma curves when your goal is to preserve as much of the source dynamic range for later color correction.

Some imaging basics:  Light hitting the sensor and the sensor’s response to that light, is effectively linear (not the incorrect use of linear to describe video gamma that still seems to be popular.) Linear has the property that as light doubles (increasing one stop), its sensor value doubles.  With an ideal 12-bit sensor, ignoring noise, there are 4096 values of linear light.  After the first detectable level of light brings our ideal sensor from 0 to 1, a doubling of light goes from 1 to 2, and the next stop from 2 to 4, and so on to produce this series 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048 and 4095 of doubling brightness (to the point where the sensor clips.) An ideal 12-bit sensor has a theoretical maximum of 12-stops of dynamic range.  If we were storing this 12-bit data as uncompressed, this is the most flexible data set (for color correction), yet this would be over 1000Mbits/s compared with today’s standard mode 1080p30 mode on HERO2 at 15Mb/s – think how fast your SD card would fill, if it could actually support that fire hose of data. Fortunately it turns out that linear is a very inefficient way of presenting light when humans are involved, as we see brightness changes logarithmically--a stop change is the same level of brightness change to us, whether it is from linear levels 1 to 2 or from 1024 to 2048.  As a result, most cameras map their sensor’s 12, 14, 16-bit linear image, to an 8, 10 or 12-bit output with a log or gamma curve, exploiting that we humans will not notice.  Even the uncompressed mode of the new Blackmagic camera maps its 16-bit linear output and only stores 12-bit with a curve – this is not lossless, but you will not miss it either. Lossless versus lossy is an argument you might have heard me present before, to the same conclusions.

If we remained in linear, converting to 8-bit from 12-bit would truncate the bottom 4 stops of shadows detail, we will notice that.  So a conventional 2.2 gamma curve does the following with its mapping (top 5 stops shown.)
12-bit Linear input
8-bit Gamma 2.2 output
Codes per stop
256
73
19
512
100
27
1024
137
37
2048
187
50
4095
255
88

So gamma curves don’t fully embrace a human visual model, with many more codes used in the brightest stop as compared with the darker stops.  The perfect scenario might be to have the 256 codes divided amongst the usable stops, e.g. 11 stops would be around 23 codes per stop.  Remember, this is for an ideal sensor (i.e. noise free) and this is not going to happen.  The darkest usable stop is mostly noise, whereas the brightest stop is mostly signal, we need a curve to handle the allocation of our code words with this in mind.  

The top 5-stops of the Protune log curve:
12-bit Linear input (idealized)
8-bit Protune output
Codes per stop
256
112
33
512
146
34
1024
181
35
2048
218
37
4095
255
37

While the darkest useable stop have a similar number of code words as the gamma curve, Protune distributes the codes are more evenly over the remaining stops, more code-words are reserved for shadow and mid-tone information. 

While I glossed over this before, again why not just have 23 code words per stop?  This has to do with compression and noise.  Noise is not compressible, at least without it looking substantially different than this input, and the compressor, H.264, CineForm or any other codec, can’t know signal from noise.  So if too many code words represent noise, quality or data-rate has to give.  The Protune curve shown above will produce smaller files, and generally be more color correctable than using fixed code words per stop. We have determined the best curve to preserve dynamic range without wasting too much data to preserve noise.

Side note for other RAW cameras: We have extended our knowledge gained while developing to the Protune curve to calculating of the best log curve for a particular dynamic range. This feature has now been included in the commercial version of CineForm Studio (Windows versions of Premium and Professional), so that the RAW camera shooter, such as from Canon CR2 time-lapse videography, to Blackmagic CinemaDNG files, can optimize the log encoding of their footage.  Of course transcoding to CineForm RAW at 12-bit rather than 8-bit H.264 helps greatly, yet the same evening out of the code-words per stop to applies as it does in the HERO2 camera running Protune.

Protune couldn’t exist as just a log curve applied upon the existing HERO2 image processing pipeline, we had to increase the bit-rate so that all the details of the wider dynamic range image could be preserved. But we didn’t stop there.  As we tuned the bit-rate, we also tweaked the noise reduction and sharpening, turning both down so that much more natural detail is preserved before compression is applied (at a higher data rate required to support more detail.) Automatically determining what is detail and what is noise, is a very difficult problem, so delaying more of these decisions into post allows the user to select the level noise reduction and sharpening appropriate to their production.  I personally do not apply post noise reduction, happy working Protune as it comes from the camera, adding sharpening to taste.  

The CineForm connection:  35Mb/s H.264 H264 is hard to decode, much harder than 15Mb/s. So transcoding to a faster editing format certainly helps, and that comes for free with GoPro CineForm Studio software.  Also, the new Protune GoPro clips carry metadata that CineForm Studio detects and automatically develops to look more like a stock GoPro mode, cool-looking and ready for show.  All these changes are stored as CineForm Activate Metadata, are non-destructive and reversible, all controlled with the free CineForm Studio software.  GoPro is working to get professional features in the hands of the everyday shooter, and the CineForm codec and software is an increasing part of that solution.  

There is so much to this story, but I’m sure I’ve gone on too long already. Thank you for reading.

P.S. Sorry for the lack of sample images, Protune launched while I'm on vacation, my internal connection is way limited at the moment. 

---

Added sample images in the next blog  Why I Shoot Protune -- Always!