PatrickGSR94

Been using a GoPro on my commutes for a couple of months now. It's a Hero4 Silver and I run it in the wide angle setting, 1080p, 60 FPS which seems to make for smoother video.

However when viewing the video, even before any type of processing or editing or YouTube uploading (watching the file from the computer using VLC), it seems like I can't ever make out details like license plates of cars passing me.

I don't understand it because I've seen thousands of videos from commuter cyclists in the UK and their cars' number plates always show up very clearly in videos. When I look at my videos in VLC and pause a frame, every single frame with anything moving (car passing, etc) looks very blurry. When I press the frame-by-frame advance button, it appears to advance a frame only every time I click it, and each frame almost looks like a blend of two frames.

So what's the deal? Is my camera not shooting true 60 fps? Is there some kind of setting I'm missing that's causing objects to not appear sharp in the video when paused? Or should I change it to 30 fps instead of 60?

PatrickGSR94

Also I should mention that I'm using a K-Edge bar mount, which keeps the camera MUCH more steady than the GoPro handlebar mount. But even with that it seems like the image isn't clear enough to make out license plates.

canklecat

Vibration is the detail killer. Mechanical vibration reduction (lens or sensor stabilization) or optical image stabilization are better than software stabilization. Simple software stabilization tends to lock onto the horizon or large geometric shapes like buildings, at the expense of desired detail, and the results are often swimmy and almost nauseatingly disorienting to watch.

For camera/lens systems that lack mechanical stabilization, the best and simplest stabilization method is to mount the camera in a way that isolates the camera from vibration. Often the results are better than sensor/lens stabilization.

The best, most stable and detail-rich bicycling videos I've seen were taken with chest harness mounts. Helmet mounts aren't bad. Handlebar and frame mounts are the worst, by far.

I've been experimenting with homebrewed handlebar mounts to find one that works but the simplest physical device that isolates the camera from bike vibration -- foam or gel -- is hard to stabilize and results in videos with a lot of sway.

However my most recent experiment with a handlebar mount for my Nikon V1 delivered enough real detail to identify license plates. But it's a kludge and needs work to be practical. I used heavy duty Velcro, which is thick enough to offer some vibration isolation, but allows a lot of sway. I ended up using medium duty zip ties to secure the mount, but it shifted too much on rough pavement and light off-road paths, so it's not really practical yet.

The physical mount itself is an ugly kludge, a heavy metal baseplate adapter from my old Nikon F3 and MD-4 motor drive, with a nylon ball head. It helped isolate the camera from vibration and enabled adjusting the camera to the desired angle and level, but the height of the rig above the handlebar exaggerated every minor turn of the bike, so the horizon sways too much.

Really, the simplest physical solution is a chest harness mount. Check out the earlier videos by Bike Blogger on YouTube (before he recently added a rear facing frame mounted video camera).

However keep in mind that YouTube videos suffer from compression which robs fine detail, even in 720 or 1080 HD. Vimeo and other sites might be better platforms if fine detail is needed. I'm working on a safety oriented video that depends on seeing fine detail in slow motion and YouTube is useless for that. I've done some test uploads and had friends watch to get objective feedback, and the YouTube videos just can't hack it.

PatrickGSR94

Well I compare to other YouTube channels like CycleGaz, who seems to have exceptionally clear video with lots of detail, even on his bar mounted ones. All his recent videos the past few months are bar-mounted as he got a new helmet that won't work with the GoPro helmet mount pads (same with my helmet, can't use the pad mounts due to the vents).

Compare a video like this, which looks amazingly clear:

To one of my recent videos, which looks just so-so, and most of the time I can't make out the license plates at all of cars passing by me:

Could it be that UK number plates are simply more legible than US plates? Or is it because the cars over there are often passing the camera closer than the cars passing my camera?

canklecat

Okay, I took a look at about 10 second of each video at 1080, which is a chore on my slowpoke DSL (no broadband available in my area). I don't see any significant differences in fine detail between your video and the UK cyclist's.

The main factors?

1. UK license plate alpha-numerics are huge compared with ours, and with better contrast between letters/numbers and background. When we're dealing with pixel level detail that's a huge factor.
2. Lens performance. Very underrated in the action cam field. I'll explain in a moment.
3. In-camera processing, both manufacturer-determined and beyond user control, and user-selected options.
4. Editing software and file format output.

Regarding 1 -- Take a look at the CycleGaz video fine details along the edges - they're no better than yours, or mine using a Nikon V1 at 720p. Compare comparably sized letters, numbers and background contrast, including power lines, etc., and the fine details are very comparable.

Regarding 2 & 3 -- Most tiny sensor digicams (still and video) fitted with compact lenses that don't protrude far demand some compromises. The real optical performance usually is like a fisheye lens. In-camera software correction to produce straight edges will soften fine detail at the edges. That's unavoidable. The best true detail will be at the center, so aim the camera where you expect to need fine resolution.

Regarding 3 -- In-camera adjustments (preset and user-defined) for sharpening and contrast can make a huge difference in resolution of fine detail. High sharpening and contrast may produce superficially "sharper" looking photos and videos, but the compromises usually involve halos around high contrast areas such as black text against a white or light colored background. Which brings us to...

4 -- Editing. Most free editing software like Windows Movie Maker, Avidemux and YouTube's own will all rob fine detail and impose compression that robs fine detail. To avoid that we'd need to shoot in camera raw, use an editor that can handle camera raw, and output at the least lossy output format that's also practical for online use. But even then YouTube will impose additional compression that robs fine detail.

An example... those halos around high contrast areas I mentioned? Lossy compression will average together the black alpha-numerics, the white background and the halo area. The resulting mush will make it difficult to discern the difference between the capital letter B and the numeral 8.

Colors in fine detail suffer as well. For example, a video I recorded this weekend demands that a tiny spot of red be visible or the video is useless. The spot of red covers only a few pixels as my camera recorded it. Even in the unedited .mov file it's visible only in a few frames. But after editing in Windows Movie Maker, or Avidemux, the spot of red is rendered as light gray. And after YouTube compression does its dirty deeds, the spot is invisible. Without that red spot the purpose of the demo is lost, it's useless. So the only way to make the demo practical is to enlarge from a screen cap of the original .mov file, use a better video editor, and upload to Vimeo or other site that preserves better fidelity.

One reason I'm experimenting with homebrewed mounts is in hopes of finding a compromise between a rigid mount that ensures metal to metal contact (which seems logical) and a mount that reduces vibration. So I'm experimenting with nylon mounts and ball heads, various cushioning materials, etc. The resulting videos do seem to preserve more fine detail, but there's more sway than some viewers might find acceptable. And part of the improvement may be attributable to Nikon's Vibration Reduction lenses. I don't know whether any currently available action cams offer mechanical vibration reduction or optical stabilization in the lens optical elements or sensor.

Most cyclists wouldn't be happy with even the smallest Nikon cameras/lenses using VR, Canons using lens based optical image stabilization, or Olympus or other Micro 4:3 cameras using sensor based stabilization. Those cameras are large and heavy compared even with the GoPros which are no longer the smallest, lightest action cams. My Nikon V1 is large and heavy for a bicycle mounted camera, and probably won't withstand too much jolting from handlebar mounting.

canklecat

Incidentally, regarding your video, I see that stuff every day, every ride. I just don't even pay attention anymore, and don't let it ruffle my feathers. I don't bother signalling people about road hazards or oncoming traffic because it doesn't help and they'll be confused at best, and at worst think I'm making obscene gestures in some alien sign language. I don't even care if they yell at me to get off the road or cuss. I just smile and wave.

If they don't crowd me off the road or pass me unsafely, I don't worry about it. If they endanger themselves or other drivers, there's nothing I can do about that. As long as it doesn't affect my safety, I'm not going to waste energy worrying about it.

I hope this doesn't sound smug or fatalistic. Thirty years ago when I cycle commuting daily I reacted pretty much the same way as most of those commuter cycling videos I see on YouTube, with the notable exception of Bike Blogger, who's practically a Zen master of inner peace and contentment on the road. Amazing for a fellow who's probably in his early 20s. At his age I was furious most days in bicycle commuting.

Then, 15 years ago, I went through a period of serious PTSD after being struck and nearly crippled by a reckless driver, while I was driving with my grandsons. Fortunately the kids were okay, but I still experience chronic and occasionally severe neck pain and headaches from a permanently splintered C2 vertebrae. For a couple of years I couldn't drive, and even after that I experienced tunnel vision and road rage at every tiny mistake made by other drivers. I went into counseling and worked hard to overcome that.

When I resumed cycling last year I decided it would work only if I took an entirely different attitude, and worked on that every time I rode in public. I intend to enjoy myself and choose not to react to every little thing that happens on the road. I'll reserve my energy for the most egregious, direct and deliberate or negligent recklessness.

And I try to remind myself of the majority of drivers who are courteous and even helpful. I do a lot of smiling and waving at folks who obviously choose to give me plenty of time to pass, even when it means they miss an opportunity to scoot out of the parking lot or intersection into a break in rush hour traffic.

PatrickGSR94

So I switched to the Medium lens setting instead of wide, and now license plates and other things do appear a little more clear I guess because they're larger in the video.

lhendrick

Good move going from Wide angle to medium. Much better results in general.

canklecat

Could be. Most action cams have relatively few pixels for resolving fine detail such as license plates, so they can use all the help they get. And the lens might resolve better at that setting, although I'm not familiar with the GoPro lens and don't know whether any optical elements actually move or it's just a crop within the same focal length.

FWIW, since this discussion began I've been experimenting with various video settings and mounts. I'm seeing the best detail at 720/60 -- much better than 1080/30 or 1080/60i. Better chance of capturing a sharp frame between the vibration and motion blur. So any dedicated action cam I might buy will need to handle a true 1080/60 or better (not interlaced, interpolated, etc.).

And my handlebar mount was much sharper than the headtube mount. Too much vibration and road noise on the head tube mount made the videos useless -- only advantage was a more stable horizon and less "swimmy" look. The best handlebar mount position was directly over the stem, but that required a farm-rigged bunch of zip ties that were too fragile -- one snapped as I got home, just from holding the stem while wheeling the bike inside. No camera mounted, no harm done.

I'm not sure my camera/lens vibration reduction made as much difference as I'd initially hoped. Most mechanical stabilization is intended to reduce the effect of handheld motion, and seems to be beyond its limits in handling road induced vibration.

I'll keep testing after this week's bout of rain stops.

PatrickGSR94

I use the K-Edge mount that puts the camera right above the bars, on the right side of the stem. I'm starting to think the K-Edge Out Front mount might be better. I have 105-5700 brifters with all the cables coming out from under the bar tape, so it makes it a little difficult to get the camera mounted in a good spot.

I should do a comparison of shots between the K-Edge mount and the GoPro handlebar mount. The difference is staggering.