Hi,
I need advice on setting the CRF for home videos. Until now, I’ve been shooting in FHD/50p h.264 and exporting with CRF17. I’m not sure about the preset since I’ve never focused on it—it was the default in Voukoder for Adobe Premiere. Now, I’m also shooting with a Pixel 8 Pro. The videos are 4K/60p and h.265. Sometimes I shoot in HDR, which also has 10-bit color. What CRF equivalent should I choose for these two types of videos to match the original export?
I assume that CRF 17 is indistinguishable from the original, yet still saves space. I also know it's stated that h.265 can increase the CRF by 2 points while maintaining the same quality. And also by about 2 points up if I choose the preset 'Very Slow' instead of 'Medium.' However, I am unclear about what CRF set for 4K/60p h.265 or the same+10-bit
Thank you.
+ Reply to Thread
Results 1 to 15 of 15
-
-
The goal of a constant rate factor is to (more or less) "guarantee" a maximum for a loss of quality. That should work with only little dependency on resolution or frame rate, as it calculates a difference between the input and the reconstructed output. What will differ will be the required bitrate to achieve a loss below the given threshold. What may also differ may be the kind of loss, depending on the allowed features to save bitrate, artifacts which may appear more or less obvious to your brain. IMHO, the range to "transparent quality" will be between 15 and 18 in modern psycho-visually optimized codecs (smaller CRF = less loss = more space required).
Other users may have a different opinion. Please be patient and decide after more replies. -
Note that the "very slow" preset is really slow, so it's your choice.
As for me, for 4K you can easily increase the CRF by 2 (to 19).
Pixel records in VFR, so either you accept a high probability of duplicate frames, or you need to process the material in another program first (e.g. VDub+Avisynth). Your Pixel is probably recording in Full Range, so you also need to take into account that Premiere is seriously lacking in this area.
P.S. I tried switching from Vegas to Premiere. Premiere really sucks -
Thank you for info. Is there any way how to take videos on Pixel 8pro CFR 60p? For example with OpenCamera? I tried set in OC 60fps, but it is still VFR(medianfo).
-
One of the reasons you can get away with encoding HD or UHD video at a lower quality than SD, is the more the video is upscaled for viewing on a UHD TV, the more any encoding artefacts are upscaled too. Here's some screenshots from an old thread when I was involved in a similar discussion. The source was a UHD video encoded using CRF 24. I downscaled it to 900x400 and encoded it twice, once at CRF 24 and again at CRF 18. In this case the x264 encoder was used but I cant imagine why the same wouldn't apply when using x265. Maybe even more so, especially at 4k. All the screenshots are taken with the video running full screen on a 1080p display.
4k source displaying on a 1080p monitor
https://forum.videohelp.com/attachments/54812-1599500597/UHD%20at%201080p%20and%20CRF24.png
Downscaled and encoded at CRF 18
https://forum.videohelp.com/attachments/54814-1599500765/SD%20at%201080p%20and%20CRF18.png
Downscaled and encoded at CRF 24
https://forum.videohelp.com/attachments/54813-1599500597/SD%20at%201080p%20and%20CRF24.png
I generally encode SD at CRF 18, HD at CRF 19 and 1080p or higher at CRF 20, and that's whether I'm using x264 or x265, but if space is an issue, you can get away with a lower quality at higher resolutions. I don't do much encoding of video over 1080p myself though.Avisynth functions Resize8 Mod - Audio Speed/Meter/Wave - FixBlend.zip - Position.zip
Avisynth/VapourSynth functions CropResize - FrostyBorders - CPreview (Cropping Preview) -
More modern codecs than AVC (H.264), like HEVC (H.265), VP9/10, VVC (H.266) and AOM AV1 also have features to compress large scale content with little details even more efficiently. That may cause some more obvious blurring, though. Therefore the SAO feature (Sampling Adaptation Offset) in HEVC was commonly nicknamed "smooth all out" while it was a bit too powerful in x265.
-
I did some measuring by vmaf metric and original video FHD/50p from camera Sony AX53 coded ffmpeg. Video has 31s. Processor Ryzen 7 5700X. Value on the end of each row is VMAF average score. HEVC does not make sense in any circumstance. Takes much longer for coding and vmaf is worse then AVC.
C.mp4: original-31s 200MB 100.00
HEVC_17_slow_151s_230MB.mp4: 99.88
HEVC_17_slower_16m_237MB.mp4: 99.87
HEVC_18_slow_143s_202MB.mp4: 99.82
AVC _17_slower_58s_ 174MB.mp4: 99.79
AVC _17_medium_20s_174MB.mp4: 99.76
HEVC_19_slow_137s_176MB.mp4: 99.72
HEVC_17_medium_181s_214MB.mp4: 99.72
HEVC_19_slower_14m_ 184MB.mp4: 99.70
HEVC_20_slow_135s_153MB.mp4: 99.60
HEVC_20_slower_13m_ 160MB.mp4: 99.57
AVC _19_slower_55s_ 133MB.mp4: 99.55
AVC _19_medium_20s_134MB.mp4: 99.49
HEVC_19_medium_62s_161MB.mp4: 99.45
AVC _20_slower_54s_ 116MB.mp4: 99.38
AVC _20_medium_20s_116MB.mp4: 99.29
HEVC_20_medium_59s_139MB.mp4: 99.26
HEVC_23_slow_111s_97MB.mp4: 99.05
HEVC_23_slower_11m_ 99MB.mp4: 99.05
AVC _23_slower_47s_ 73MB.mp4: 98.53
HEVC_23_medium_49s_ 87MB.mp4: 98.36
AVC _23_medium_17s_ 72MB.mp4: 98.32
HEVC_28_slow_87s_42MB.mp4: 96.88
HEVC_28_slower_8m_ 44MB.mp4: 97.09
HEVC_28_medium_38s_ 36MB.mp4: 93.83
AVC _28_slower_41s_ 33MB.mp4: 93.73
AVC _28_medium_15s_ 32MB.mp4: 93.02
I properly checked videos on 27in 2K monitor. CRF28 AVC slowest/medium and CRF28 HEVC medium is visible worse than original.
HEVC_28_slower is worse only around people when run.
AVC CRF20 slower is almost ok. And medium is acceptable, because it is 2,5x faster. But it is around runninng people liiittle bit worse than slower.
But when i coded 13s 4K/30p from Pixel 8pro 72MB it is different. HEVC is much better. What is really good, the worst AVC CRF34 medium is aprox.5x smaller and it is almost impossible recognize it is worse then original. But closer searching... it is visible. But HEVC CRF28 medium is not even from 30cm distance.
AVC and HEVC produce larger file then original when CRF23 and smaller is used. What is intersting, vmaf is 71 and is ok. FHD has 99 and it is not ok.
C.mp4: 99.99
HEVC_17_slower_ 11m_159MB.mp4: 76.09
HEVC_17_medium_53s_146MB.mp4: 75.91
HEVC_19_slower_ 10m_129MB.mp4: 75.90
HEVC_20_slower_ 9m_115MB.mp4: 75.74
HEVC_19_medium_49s_117MB.mp4: 75.56
HEVC_21_slower_ 9m_102MB.mp4: 75.55
HEVC_20_medium_48s_104MB.mp4: 75.27
HEVC_23_slower_ 8m_80MB.mp4: 74.88
HEVC_21_medium_ 44s_92MB.mp4: 74.88
HEVC_23_medium_ 41s_71MB.mp4: 73.63
HEVC_28_slower_ 6m_ 40MB.mp4: 71.06
HEVC_28_medium_31s_34MB.mp4: 68.05
AVC _17_slower_ 63s_153MB.mp4: 54.57
HEVC_17_slow_145s_185MB.mp4: 54.51
AVC _17_medium_21s_162MB.mp4: 54.41
HEVC_18_slow_141s_168MB.mp4: 54.38
AVC _19_slower_59s_123MB.mp4: 54.29
HEVC_19_slow_136s_151MB.mp4: 54.24
AVC _20_slower_57s_111MB.mp4: 54.10
HEVC_20_slow_130s_136MB.mp4: 54.06
AVC _19_medium_20s_130MB.mp4: 54.05
AVC _21_slower_55s_99MB.mp4: 53.86
AVC _20_medium_19s_115MB.mp4: 53.81
AVC _21_medium_19s_102MB.mp4: 53.53
HEVC_23_slow_112s_97MB.mp4: 53.24
AVC _23_slower_51s_78MB.mp4: 53.21
AVC _23_medium_18s_80MB.mp4: 52.77
HEVC_28_slow_89s_49MB.mp4: 50.64
AVC _28_slower_42s_39MB.mp4: 50.18
HEVC_30_slower_6m_37MB.mp4: 49.64
AVC _28_ medium_15s_40MB.mp4: 49.55
HEVC_30_slow_84s_37MB.mp4: 49.24
AVC _30_ slower_40s_29MB.mp4: 48.24
HEVC_32_ slower_6m_27MB.mp4: 48.06
AVC _30_ medium_15s_29MB.mp4: 47.58
HEVC_30_ medium_39s_32MB.mp4: 47.51
HEVC_32_slow_78s_27MB.mp4: 47.61
HEVC_34_ slower_5m_19MB.mp4: 46.15
AVC _32_ slower_39s_21MB.mp4: 45.80
HEVC_32_ medium_38s_24MB.mp4: 45.39
AVC _32_ medium_15s_21MB.mp4: 45.15
HEVC_34_slow_72s_19MB.mp4: 45.67
HEVC_34_medium_37s_17MB.mp4: 42.86
AVC _34_ slower_ 36s_15MB.mp4: 42.84
AVC _34_ medium_14s_15MB.mp4: 42.15Last edited by Divok; 6th May 2025 at 18:48.
-
I recorded a 4K/25p clip with the Sony AX-53. It's 10 seconds long and about 80 MB in size. It should be a much better source, since the camcorder has a larger sensor and better optics. The VMAF score reflects that – it's noticeably higher.
From what I can see, HEVC is not twice as efficient as AVC. With HEVC, the slower preset actually produces a larger file than the medium preset. Interestingly, it's the opposite for AVC – the slower preset gives a smaller file than medium.
In general, HEVC produces better 4K quality at the same CRF, but usually results in a larger file compared to AVC.
Number in front of preset means CRF.
D.mp4 - 80MB 99.99
HEVC_17_slow_102s_91MB.mp4 99.99
HEVC_17_slower_10m_91MB.mp4 99.99
HEVC_18_slow_97s_81MB.mp4 99.99
AVC_17_medium_16s_94MB.mp4 99.98
AVC_17_slower_45s_81MB.mp4 99.98
HEVC_19_slow_91s_71MB.mp4 99.98
HEVC_20_slow_86s_62MB.mp4 99.98
HEVC_23_slow_69s_38MB.mp4 99.98
AVC_19_medium_15s_72MB.mp4 99.98
AVC_19_slower_41s_65MB.mp4 99.98
AVC_20_medium_14s_62MB.mp4 99.98
AVC_20_slower_39s_57MB.mp4 99.98
HEVC_17_medium_46s_88MB.mp4 99.98
HEVC_19_medium_40s_66MB.mp4 99.98
HEVC_19_slower_8m_72MB.mp4 99.98
HEVC_20_medium_39s_56MB.mp4 99.98
HEVC_20_slower_8m_63MB.mp4 99.98
HEVC_23_slower_6m_39MB.mp4 99.98
AVC_23_slower_34s_34MB.mp4 99.95
AVC_23_medium_13s_36MB.mp4 99.94
HEVC_23_medium_31s_33MB.mp4 99.93
HEVC_28_slow_50s_14MB.mp4 99.37
HEVC_28_slower_230s_15MB.mp4 99.34
HEVC_30_slow_46s_10MB.mp4 98.48
HEVC_30_slower_201s_10MB.mp4 98.46
HEVC_28_medium_24s_12MB.mp4 98.23
AVC_28_slower_28s_13MB.mp4 97.83
AVC_28_medium_11s_13MB.mp4 97.52
HEVC_32_slow_43s_7MB.mp4 96.68
HEVC_32_slower_179s_7MB.mp4 96.78
HEVC_30_medium_22s_8MB.mp4 95.83
AVC_30_slower_27s_9MB.mp4 94.45
AVC_30_medium_10s_9MB.mp4 93.56
HEVC_34_slow_41s_5MB.mp4 93.23
HEVC_34_slower_164s_5MB.mp4 93.57
HEVC_34_medium_21s_4MB.mp4 86.88
AVC_32_slower_26s_7MB.mp4 89.00
AVC_32_medium_10s_7MB.mp4 88.00
AVC_34_slower_25s_5MB.mp4 82.53
AVC_34_medium_10s_5MB.mp4 81.44Last edited by Divok; 6th May 2025 at 17:26. Reason: Added preset slow
-
I can test it. No problem. What video are you interest abou? I will add it
Last edited by Divok; 6th May 2025 at 06:38.
-
These are my standard h265 settings. I'm not saying they're perfect, or even for everyone, but if you have time to kill I'd be interested to know how they compare to your previous tests for quality. Thinking about it, with psy increased a little, it might do worse according to a quality metric.
--crf 18.0 --profile main10 --preset slow --deblock=-1:-1 --no-open-gop --ctu 32 --rc-lookahead 40 --lookahead-slices 0 --rd 4 --psy-rd 2.0 --rdoq-level 2 --psy-rdoq 1.5 --aq-strength 1.3 --no-strong-intra-smoothing --no-sao --sar 1:1
Or for animation
--crf 18.0 --profile main10 --tune animation --preset slow --deblock=0:0 --no-open-gop --rc-lookahead 40 --lookahead-slices 0 --rd 4 --psy-rd 1.0 --rdoq-level 2 --psy-rdoq 1.0 --aq-strength 1.0 --no-sao --sar 1:1
Unless you have a device that doesn't support it, although it's been part of the spec from the beginning for HEVC, I wouldn't bother with 8 bit encoding. 10 bit should help nudge the quality up a tad, assuming all other settings are the same. Did your tests include any 10 bit encodes? If not it might with worth a try.Avisynth functions Resize8 Mod - Audio Speed/Meter/Wave - FixBlend.zip - Position.zip
Avisynth/VapourSynth functions CropResize - FrostyBorders - CPreview (Cropping Preview) -
I have no 10bit video and encoded videos are 8bit.
I added preset slow. And HEVC_X_ is with hello_hellos setting "--crf 18.0 --profile main10 --preset slow --deblock=-1:-1 --no-open-gop --ctu 32 --rc-lookahead 40 --lookahead-slices 0 --rd 4 --psy-rd 2.0 --rdoq-level 2 --psy-rdoq 1.5 --aq-strength 1.3 --no-strong-intra-smoothing --no-sao --sar 1:1"
4K/30p 13s 72MB 8bit vmaf:100.00
HEVC_17_slower_11m_159MB.mp4: 76.09
HEVC_17_medium_53s_146MB.mp4: 75.91
HEVC_19_slower_10m_129MB.mp4: 75.90
HEVC_20_slower_9m_115MB.mp4: 75.74
HEVC_19_medium_49s_117MB.mp4: 75.56
HEVC_21_slower_9m_102MB.mp4: 75.55
HEVC_20_medium_48s_104MB.mp4: 75.27
HEVC_23_slower_8m_80MB.mp4: 74.88
HEVC_21_medium_44s_92MB.mp4: 74.88
HEVC_23_medium_41s_71MB.mp4: 73.63
HEVC_28_slower_6m_40MB.mp4: 71.06
HEVC_28_medium_31s_34MB.mp4: 68.05
HEVC_X_17_slow_143s_185MB.mp4: 54.64
AVC_17_slower_63s_153MB.mp4: 54.57
HEVC_X_18_slow_140s_168MB.mp4: 54.52
AVC_17_medium_21s_162MB.mp4: 54.41
HEVC_X_19_slow_135s_152MB.mp4: 54.38
AVC_19_slower_59s_123MB.mp4: 54.29
HEVC_X_20_slow_131s_138MB.mp4: 54.21
AVC_20_slower_57s_111MB.mp4: 54.10
AVC_19_medium_20s_130MB.mp4: 54.05
HEVC_X_21_slow_127s_124MB.mp4: 54.01
AVC_21_slower_55s_99MB.mp4: 53.86
AVC_20_medium_19s_115MB.mp4: 53.81
AVC_21_medium_19s_102MB.mp4: 53.53
HEVC_X_23_slow_118s_99MB.mp4: 53.43
AVC_23_slower_51s_78MB.mp4: 53.21
AVC_23_medium_18s_80MB.mp4: 52.77
HEVC_X_28_slow_96s_52MB.mp4: 50.80
AVC_28_slower_42s_39MB.mp4: 50.18
HEVC_30_slower_6m_37MB.mp4: 49.64
AVC_28_medium_15s_40MB.mp4: 49.55
HEVC_X_30_slow_88s_39MB.mp4: 49.41
HEVC_32_slower_6m_27MB.mp4: 48.06
AVC_30_slower_40s_29MB.mp4: 48.24
HEVC_X_32_slow_83s_29MB.mp4: 47.80
AVC_30_medium_15s_29MB.mp4: 47.58
HEVC_30_medium_39s_32MB.mp4: 47.51
HEVC_34_slower_5m_19MB.mp4: 46.15
HEVC_X_34_slow_79s_21MB.mp4: 45.89
AVC_32_slower_39s_21MB.mp4: 45.80
HEVC_32_medium_38s_24MB.mp4: 45.39
AVC_32_medium_15s_21MB.mp4: 45.15
HEVC_34_medium_37s_17MB.mp4: 42.86
AVC_34_slower_36s_15MB.mp4: 42.84
AVC_34_medium_14s_15MB.mp4: 42.15
FHD/50p 8bit 200MB 31s vmaf:100
HEVC_X_17_slow_142s_237MB.mp4: 99.91
HEVC_17_slow_151s_230MB.mp4: 99.88
HEVC_17_slower_16m_237MB.mp4: 99.87
HEVC_X_18_slow_139s_209MB.mp4: 99.85
HEVC_18_slow_143s_202MB.mp4: 99.82
AVC _17_slower_58s_ 174MB.mp4: 99.79
HEVC_X_19_slow_132s_183MB.mp4: 99.76
AVC _17_medium_20s_174MB.mp4: 99.76
HEVC_19_slow_137s_176MB.mp4: 99.72
HEVC_17_medium_181s_214MB.mp4: 99.72
HEVC_19_slower_14m_ 184MB.mp4: 99.70
HEVC_X_20_slow_126s_160MB.mp4: 99.65
HEVC_20_slow_135s_153MB.mp4: 99.60
HEVC_20_slower_13m_ 160MB.mp4: 99.57
HEVC_X_21_slow_118s_139MB.mp4: 99.52
AVC _19_slower_55s_ 133MB.mp4: 99.55
AVC _19_medium_20s_134MB.mp4: 99.49
HEVC_19_medium_62s_161MB.mp4: 99.45
AVC _20_slower_54s_ 116MB.mp4: 99.38
AVC _20_medium_20s_116MB.mp4: 99.29
HEVC_20_medium_59s_139MB.mp4: 99.26
HEVC_23_slow_111s_97MB.mp4: 99.05
HEVC_23_slower_11m_ 99MB.mp4: 99.05
HEVC_X_23_slow_106s_103MB.mp4: 99.17
AVC _23_slower_47s_ 73MB.mp4: 98.53
HEVC_23_medium_49s_ 87MB.mp4: 98.36
AVC _23_medium_17s_ 72MB.mp4: 98.32
HEVC_X_28_slow_88s_46MB.mp4: 97.29
HEVC_28_slow_87s_42MB.mp4: 96.88
HEVC_28_slower_8m_ 44MB.mp4: 97.09
HEVC_X_30_slow_81s_32MB.mp4: 95.50
HEVC_28_medium_38s_ 36MB.mp4: 93.83
AVC _28_slower_41s_ 33MB.mp4: 93.73
AVC _28_medium_15s_ 32MB.mp4: 93.02
HEVC_X_32_slow_74s_23MB.mp4: 92.55
HEVC_X_34_slow_69s_16MB.mp4: 87.67
4K/25 80MB 10s 8bit vmaf:100
HEVC_17_slow_102s_91MB.mp4 99.99
HEVC_17_slower_10m_91MB.mp4 99.99
HEVC_X_17_slow_104s_99MB.mp4 99.99
HEVC_18_slow_97s_81MB.mp4 99.99
HEVC_X_18_slow_101s_89MB.mp4 99.99
AVC_17_medium_16s_94MB.mp4 99.98
AVC_17_slower_45s_81MB.mp4 99.98
HEVC_19_slow_91s_71MB.mp4 99.98
HEVC_X_19_slow_97s_79MB.mp4 99.98
HEVC_20_slow_86s_62MB.mp4 99.98
HEVC_X_20_slow_92s_69MB.mp4 99.98
HEVC_X_21_slow_86s_60MB.mp4 99.98
HEVC_23_slow_69s_38MB.mp4 99.98
HEVC_X_23_slow_80s_44MB.mp4 99.98
AVC_19_medium_15s_72MB.mp4 99.98
AVC_19_slower_41s_65MB.mp4 99.98
AVC_20_medium_14s_62MB.mp4 99.98
AVC_20_slower_39s_57MB.mp4 99.98
HEVC_17_medium_46s_88MB.mp4 99.98
HEVC_19_medium_40s_66MB.mp4 99.98
HEVC_19_slower_8m_72MB.mp4 99.98
HEVC_20_medium_39s_56MB.mp4 99.98
HEVC_20_slower_8m_63MB.mp4 99.98
HEVC_23_slower_6m_39MB.mp4 99.98
AVC_23_slower_34s_34MB.mp4 99.95
AVC_23_medium_13s_36MB.mp4 99.94
HEVC_23_medium_31s_33MB.mp4 99.93
HEVC_X_28_slow_58s_17MB.mp4 99.52
HEVC_28_slow_50s_14MB.mp4 99.37
HEVC_28_slower_230s_15MB.mp4 99.34
HEVC_X_30_slow_53s_11MB.mp4 98.79
HEVC_30_slow_46s_10MB.mp4 98.48
HEVC_30_slower_201s_10MB.mp4 98.46
HEVC_28_medium_24s_12MB.mp4 98.23
AVC_28_slower_28s_13MB.mp4 97.83
AVC_28_medium_11s_13MB.mp4 97.52
HEVC_X_32_slow_49s_8MB.mp4 97.33
HEVC_32_slow_43s_7MB.mp4 96.68
HEVC_32_slower_179s_7MB.mp4 96.78
HEVC_30_medium_22s_8MB.mp4 95.83
AVC_30_slower_27s_9MB.mp4 94.45
AVC_30_medium_10s_9MB.mp4 93.56
HEVC_X_34_slow_46s_6MB.mp4 94.35
HEVC_34_slow_41s_5MB.mp4 93.23
HEVC_34_slower_164s_5MB.mp4 93.57
HEVC_34_medium_21s_4MB.mp4 86.88
AVC_32_slower_26s_7MB.mp4 89.00
AVC_32_medium_10s_7MB.mp4 88.00
AVC_34_slower_25s_5MB.mp4 82.53
AVC_34_medium_10s_5MB.mp4 81.44Last edited by Divok; 7th May 2025 at 02:48.
-
You can encode 8 bit video as 10 bit.
The HEVC_X encodes should be 10 bit as "--profile main10" is in the command line, assuming you didn't change it.
I forgot to mention, --ctu 32 probably shouldn't be included for resolutions above 1080p as it might increase the file size/bitrate compared to the default setting.
Very short video combined with very low bitrates possibly should use 2 pass encoding instead of CRF, if the idea is to compare quality, as it's possible 2 pass will do a slightly better job distributing the bits in some extreme situations, or distribute them differently. I've compared similar encodes visually in the past (greater than CRF 30), but at the time I thought 2 pass had a very slight quality edge over CRF even though the average bitrate was almost identical. That was using x264 though, and the initial frames had a lot of detail and movement and the encoder used more bits for the keyframes and less for the others in 2 pass mode. I'm wondering if something similar could be happening with your samples as they're very short and you said there's been unexpected quality differences at times. Just a thought....Avisynth functions Resize8 Mod - Audio Speed/Meter/Wave - FixBlend.zip - Position.zip
Avisynth/VapourSynth functions CropResize - FrostyBorders - CPreview (Cropping Preview) -
Similar Threads
-
Zond 265, the first professional HEVC / H.265 Analyzer from SolveigMM
By DmitryV in forum Latest Video NewsReplies: 5Last Post: 13th May 2024, 09:02 -
CRF or CQP?
By gelo333 in forum Newbie / General discussionsReplies: 60Last Post: 31st May 2023, 07:13 -
About the value of CRF!
By Valentin98 in forum Video ConversionReplies: 10Last Post: 24th May 2022, 14:16 -
CRF values?
By Chauceratemyhamster in forum Video ConversionReplies: 18Last Post: 28th Mar 2021, 21:06 -
CRF Values
By Anakin in forum Video ConversionReplies: 11Last Post: 7th Sep 2020, 18:12