Showcase of experiments with GoPro 11 Labs settings #160
Replies: 20 comments 41 replies
-
A rather extreme low light test: https://www.youtube.com/watch?v=bAQKnCsyyX0 This was filmed in the late evening and basically complete darkness. The lights on the bike and my helmet combined with the white snow creates a challenging scene for the GoPro with an extreme dynamic range. Surprisingly, there are no blown out highlights, so it's possible to extract a lot of data from this footage. The ISO noise is excessive, partly due to the nature of the scene, but can be greatly reduced with post-processing. An interesting trade-off exemplified here is that a faster shutter (1/200) will allow the hypersmooth stabilization to work quite well, at the cost of more ISO noise. This is a cut version that has been edited in Resolve and rendered in UHD 120mbit/s before being uploaded to Youtube. It was originally filmed in 4:3 and stretched to 16:9 in post. This footage has not been graded or processed outside of the editing and stretch to 16:9. Standard settings:
Labs settings:
|
Beta Was this translation helpful? Give feedback.
-
Good stuff. The Low light test would benefit from a pass through NeatVideo. Setting NR01=1 with a high ISO, is intended for post noise reduction. There is more coming in the next Labs version to allow you to push further. I've been experimenting with -3 and -4 EV combined with LOGB=400 or similar. The EV compensation extensions are to be in the next Labs release. |
Beta Was this translation helpful? Give feedback.
-
Here is my contribution to LOGB=400 NR01=1 testing. https://www.youtube.com/watch?v=wsJKA-rr0Ko Standard settings:
Labs settings:
|
Beta Was this translation helpful? Give feedback.
-
Another set of samples, this time testing low light shadow recovery with max ISO = 800. Raw clip from camera (16:9 SuperView): https://www.youtube.com/watch?v=i-4_EW59sdQ Standard settings:
Labs settings:
My notes: On a sidenote, it seems that 60fps footage is compressed quite hard on Youtube, so the noise characteristics are somewhat camouflaged by compression artifacts. On my local computer these clips look much better. The original ungraded video file can be downloaded here in case someone wants to experiment with grading. In conclusion I learned the following from this test:
My next set of samples will focus on full frame footage in the norwegian winter forest, and after that I will look into 24fps+200mbit/s. |
Beta Was this translation helpful? Give feedback.
-
From that clip's metadata I run a subtle Neat Video denoise pass and small shadow lift to see if that helps Youtude. Only on 25s. Also run it through Vimeo for comparison. Seems 4Kp60 is really hard on these tools (with so much texture and detail.) A crude measure of information loss. The same frame extracted from the source, Vimeo and Youtude is compressed to JPEG as 1608KB, 1308KB and 1134KB respectively. So source > Vimeo > Youtube. So the down side of captioning more information, is finding a way to share that quality. |
Beta Was this translation helpful? Give feedback.
-
Went for a small stroll in the forest on sunday and gathered a few more LOGB=400 shots in varying lighting conditions. Link: https://www.youtube.com/watch?v=SXJoXiofMm4 These are full frame clips that have been cropped top and bottom to 16:9, which I seem to like the most for these kinds of shots. The footage is graded slightly with a warmer temperature and a few other color tweaks. Sorry about the squeaking, it was very cold and my backpack apparently was not too happy the temperature. Standard settings:
Labs settings:
My notes: Tomorrow the temperature is forecast to rise up to around -8 celsius, which means I can get some more footage without losing too many fingers. Stay tuned :) |
Beta Was this translation helpful? Give feedback.
-
Hi quick basic question, if I leave all camera settings as unset in labs and program in my shutter max/min and bitrate etc, as permanent I assume whenever I change between video modes the settings for labs carry across? |
Beta Was this translation helpful? Give feedback.
-
Hi, I sent an email, there is a bug in the app, if I select LOGB the code produced Is LOGO=400 the text description is also wrong, if I select LOGO I get the text description for the LOGB details in app. It appears some other items are out in the app, it’s like a new item has been added but placed all the descriptions one place out. Another issue I am struggling to understand is the shutter max and min, the following code does what I want MLOGB=400!MEXPX=100!MEXPN=900!MBITR=180!MNR01=1!MEXPS=1 but the me the expn (minimum) should actually be the max shutter not minimum? So the code is saying max shutter=100 and minimum=900. |
Beta Was this translation helpful? Give feedback.
-
The daylight is slowly coming back here in Norway and I finally got to do a short ride in the snow-covered forest trails nearby. Link: https://www.youtube.com/watch?v=DpganRr_B4w Edit: I've also published a longer version (7:30) with slightly less saturated colors: https://www.youtube.com/watch?v=qtM8OxVIsnA These are full frame clips that have been squeezed to 16:9 (similar to hyperview). The footage is graded with Davinci Resolve Studio, and I have applied some distortion correction as well as temperature correction. There is no noise reduction applied in post :) Standard settings:
Labs settings:
My notes: GP11 5.3k 60fps with noise reduction enabled and ISO 1600 There is a significant difference in ISO levels in these screenshots, but on a stock GP11 this is necessary in low light conditions to maintain hypersmooth performance. Also note that the quality of the footage is reduced by Youtube's compression, but it's still miles and miles better than the footage you can hope to achieve with a stock GP11 in low light conditions. A highly appreciated bonus is that hypersmooth works extremely well even in very poor lighting conditions with these settings. |
Beta Was this translation helpful? Give feedback.
-
sadly I have the mini since 3 days, and have not been able to run it stable: Labs settings: NR01=1 SanDisk Extreme Pro, 128GB, exFAT (tested multiple times the writespeed over the gopro command (480) yesterday i tried 3 times with 8 bit (everything worked till i stopped the recording after around 5 minutes, worked flawlessly... today I tried it with 10 bit on, any my recordings 5/10 have stopped randomly sometimes it worked for 5-6 minutes sometimes after 3 minutes it broke up.... what can i change to still have 10 bit and NR=1 (which Bitrate is reliable?) or is it maybe because of the 10 bit or because of the NR=1? I really think its the SD card or the GoPro mini itself has problems? The sad part is, I though everything works perfectly and I travel south east asia right now, so there is no Time for flying in the same spot a few times....have already missed some really pretty recordings on my last 3 days... I really want a field testet setting that never stops my recording (FPV Flying, turn the GoPro on and start recording when everything else is ready for takeoff, I am pretty sure its no overheat) do you know what is the threshold bitrate that I should minimum go for, without having Problems with my wished settings? How would I notice that the Bitrate settings are too low? What impact would that have? Thank you guys... |
Beta Was this translation helpful? Give feedback.
-
if you go for 200mbits on the gopro and it shuts off do you get a text on the qrlog.txt? maybe i can not go with 180 like everyone else because i dont use hypersmooth (so no crop at all) = more data written? or am I wrong? Because I use reesteady in Post |
Beta Was this translation helpful? Give feedback.
-
I made a short color grading comparison using Davinci Resolve on four short clips from high contrast scenes. The comparison is made between raw SooC clips and a simple color grade + distortion correction in Davinci Resolve. Link: https://www.youtube.com/watch?v=lOzD7eoLgmE I've been experimenting with various LOGB settings this winter but I keep coming back to LOGB=400 (along with EV Comp=-2.0). It just seems to give the best all-round results that are easy to grade. Looking forward to trying out the upcoming WIDE and GCMP settings to see if they can help squeeze out even more delicious data from the GP11 :) |
Beta Was this translation helpful? Give feedback.
-
Labs v2.10.70 has landed, and with that came the new WIDE setting. It's still full on winter here in my part of the world, so the landscape is rather bleak and monochrome-looking. Regardless, I went out for a small ride yesterday to capture some footage with the new WIDE setting enabled. I made sure to visit scenes with challenging lighting conditions to stress test the dynamic range. I've uploaded three videos: Ungraded: https://www.youtube.com/watch?v=XFo-vbPt1-A Saying that I graded this footage might be a stretch -- I basically just added a contrast curve, some sharpness, adjusted the temperature, and did some very minor adjustments with a hue vs. saturation curve. Looking at the footage in DaVinci, it's really quite impressive how much data is being captured with these settings. Here's a frame from the footage both ungraded and graded, accompanied by their respective waveforms:
We can see that there is a small amount of highlights being clipped. Since the shadows are a far cry away from being crushed in these scenes, I could probably add EVBS=-1 for a total EV comp of -3.0 to save more of the highlights. Modifying the LOG curve with an offset might be an option too. My initial impression of the new WIDE setting is very positive. It seems to add noticeably more headroom for color grading and artistic expression. Personally I am not very experienced with color management though, so I hope that someone more versed might share some tips or even LUT's to get the most out of these settings. Standard settings:
Labs settings:
Edit: After dorking around some more in DaVinci I found that adding a node with a color space transform seems to bring more pleasing results. Here are the settings I used: I also changed the timeline output color space to Rec.2020, but that didn't seem to make a difference. I have no idea what I'm doing now, so anyone knowledgeable about this should probably chime in :) |
Beta Was this translation helpful? Give feedback.
-
To get accurate colors you have to do proper color space transform. Fortunetly Protune Native Flat is a well know gamut since Hero 3+, you can have proper conversion LUT's for it. My recommendation is to shoot with WIDE=1, LOGB=113, NR01=1, EV=-1 / -1,5, high bitrate and use ACES in davinci resolve. |
Beta Was this translation helpful? Give feedback.
-
I've been trying out a few different combinations of LOGB values with WIDE=1, and today I found LOGB=600 and EV -1.0 to work quite well. Here's a demo clip of ungraded footage: https://www.youtube.com/watch?v=U79LgNiBmks Standard settings: Labs settings: In nice and bright conditions like today with strong highlights and dark shadows, these settings seem to work well. Grading this footage is a breeze and produces quite nice results with minimal clipping. Bumping the exposure compensation to -1.0 improves the ISO noise situation while introducing some penalty to the highlights. Here's a screenshot of a scene taken in DaVinci: Using EV=-1.0 rather than EV=-2.0 seems to be favorable with regards to ISO noise in this kind of light. GCMP also helps out a bit there in fast-moving situations. Edit: I also published a more stylized edit with a warm grading and carefully processed raw audio: https://www.youtube.com/watch?v=u9feHZ_OG_4. I think this both looks and sounds absolutely gorgeous! |
Beta Was this translation helpful? Give feedback.
-
i know it's against the company interests, but is it possible to implement LOGB, NR, WIDE functions to work with Hero 10 Labs Firmware? At least the processor is the same, the image sensor interchangeble... |
Beta Was this translation helpful? Give feedback.
-
Is that a mirrorless camera strapped to your body, or are you just So far, GCMP seems to do more or less exactly what I was hoping for -- foregoing some shutter speed and stabilization performance in exchange for lower ISO noise. With the settings I use here, some jittering is visible in the footage. But the trade-off for less noise is well worth it, at least for my use cases. Standard settings: Labs settings: |
Beta Was this translation helpful? Give feedback.
-
Custom log curves and then in DaVinci Resolve what to do? To better understand what is going on in clouds and highlights I put 2 Hero11 side-by-side on the window sill and pointed them outside. https://www.youtube.com/watch?v=ilLKbolSsRs The video shows side-by-side how (in comparison to ungraded Protune Flat)
By putting this out here I am looking for collaboration with experienced people, with the objective to publish not just videos but "newbie-friendly" DaVinci Resolve explanations. |
Beta Was this translation helpful? Give feedback.
-
The focus last weekend was again on LOGB200,-128 and how that looks with just a simple CST in DaVinci Resolve. https://www.youtube.com/watch?v=-sSp1qiezAk Bitrate 150 MBit/s, LOGB200,-128 as FujiFilm F-Log emulation, WIDE=1, EV Comp -0.5 and NR01=40 with 30 fps. Graded with just a Color Space Transform as Input Rec709 Gamm FujiFilm F-Log. A bit of noise reduction on the first 1/4 of the video, deflicker and motion blur with a clip speed of 1000. Uploaded as 4k, and a screenshot taken from YouTube as below. For the usecase of the upcoming holiday cycle tour I am happy with the quality. (Where I expect that videography will take a less prominent role in my day-to-day focus.) |
Beta Was this translation helpful? Give feedback.
-
Hello, can anyone tell me whether gopro labs is better suited for gopro 11 than for gopro 12. It seems to me that several functions of Labs do not work correctly on gopro 12 |
Beta Was this translation helpful? Give feedback.
-
Kicking off a new thread to showcase my experiments with GoPro Labs settings on my GoPro 11, with particular focus on bitrate, noise reduction, exposure settings and log curve adjustments. This thread is meant to be a source of concrete information on various labs settings and their effect on the footage. I'm happy to take requests if anyone wants me to test something in particular.
My first test was a simple ride through the city with overall excellent light conditions. The improved dynamic range is showcased around 2:09 when I point the camera towards the sun.
Raw 4:3 clip directly from camera: https://www.youtube.com/watch?v=Rtue8uhPBIs
Graded and stretched to 16:9 (i.e approximately superview): https://www.youtube.com/watch?v=boelDHDSrN8
Standard settings:
Labs settings:
As you can see, at 30fps I was able to have a stable recording for 4min22sec. The recording was stopped manually. I have not experienced any instabilities with these settings yet.
Beta Was this translation helpful? Give feedback.
All reactions