Improving Madvr HDR to SDR mapping for projector - Page 198 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 7636Likes
Reply
 
Thread Tools
post #5911 of 8106 Old 05-19-2019, 04:58 AM
AVS Forum Special Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 1,147
Mentioned: 206 Post(s)
Tagged: 0 Thread(s)
Quoted: 870 Post(s)
Liked: 1213
Quote:
Originally Posted by madshi View Post
1) Would it make sense to maybe add a separate "misses sum" line for high APL scenes and to exclude them from the other sums? Right now you're separating "false positive misses" vs "real scene misses". So there would be a 3rd line for "high APL misses". Doing that would allow us to check if switching from 35% to a higher value only helps high APL scenes, or if it could potentially help other scenes, as well. Generally, I'm not sure if we need to worry too much about high APL scenes, because Metric1 does a great job with those. On the other hand, it we can make Metric2 handle those better without any negative side effects, why not?
Just by pre-testing with various scenes at 35% and 45%, I get a lot more of false positives with 45% (meaning the value of Metric2 is more often higher during the scenes than on cut, compared to 35%).

If we don't need to worry too much about high APL scenes, why don't we try percentages lower than 35 ?

By combining both metrics, since Metric1 is always very high on high APL scenes, it is not a problem if Metric2 is a bit low, unless it is dramatically low (lower than the minimum limit to accept it as a real cut).

Quote:
Originally Posted by madshi View Post
2) A good question is how to split the testing work between you and Fer15, since you now merged all his scenes into your sheet. He had some suggestions, e.g. one of you could test false positives, the other one real scenes. Or maybe he would simply continue to test "his" test set and you would skip those and then you could throw your numbers together? It's not important to me how you split the work, of course, you can decide for yourself. Just wouldn't want you both to duplicate each other's work, to not waste any of your time.
It's always good to double check, so I don't mind doing more work.
madshi and Manni01 like this.

Last edited by Neo-XP; 05-19-2019 at 05:09 AM.
Neo-XP is offline  
Sponsored Links
Advertisement
 
post #5912 of 8106 Old 05-19-2019, 05:23 AM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
Quote:
Originally Posted by Manni01 View Post
Do we need to remeasure our files following the D3D11 native fix in build 80 [...]?
I don't think so. There might be a difference in the measurements but it should be so small that it's invisible.

Quote:
Originally Posted by Neo-XP View Post
Just by pre-testing with various scenes at 35% and 45%, I get a lot more of false positives with 45% (meaning the value of Metric2 is more often higher during the scenes than on cut, compared to 35%).

If we don't need to worry too much about high APL scenes, why don't we try percentages lower than 35 ?

By combining both metrics, since Metric1 is always very high on high APL scenes, it is not a problem if Metric2 is a bit low, unless it is dramatically low (lower than the minimum limit to accept it as a real cut).
Oh wow, I was expecting the opposite result. But that's fine. FWIW, increasing the percentage will increase *all* Metric2 results, both false positives and real scenes, so you definitely need to increase the threshold with higher percentage values (or vice versa).

So here's a new test build which now has 5% steps between 0-75%:

http://madshi.net/madVRhdrMeasure82.zip

I'd recommend to not test all of them, just pick some that seem useful. But I've still added 5% steps so if you want to fine tune, you can.
Manni01 and omarank like this.
madshi is online now  
post #5913 of 8106 Old 05-19-2019, 08:14 AM
AVS Forum Special Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 1,147
Mentioned: 206 Post(s)
Tagged: 0 Thread(s)
Quoted: 870 Post(s)
Liked: 1213
@madshi are you sure the displayed percentages are correct in madVRhdrMeasure82 ?

35% in madVRhdrMeasure82 gives me higher values than 35% in madVRhdrMeasure81.

30% in madVRhdrMeasure82 gives me the exact same values as 35% in madVRhdrMeasure81.


Last edited by Neo-XP; 05-19-2019 at 08:24 AM.
Neo-XP is offline  
Sponsored Links
Advertisement
 
post #5914 of 8106 Old 05-19-2019, 09:40 AM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
You're right, that's my fault. Do you want a new test build? The percentages are just 5% higher than they should be. So there's no 0%, it starts at 5%. Sorry about that...
Neo-XP likes this.
madshi is online now  
post #5915 of 8106 Old 05-19-2019, 11:10 AM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
Wasn't sure if you were waiting for a new test build or not, so I made one just to be safe:

http://madshi.net/madVRhdrMeasure83.zip

Same as before, only with the +5% fixed.
SamuriHL, Manni01 and Neo-XP like this.
madshi is online now  
post #5916 of 8106 Old 05-19-2019, 11:37 AM
AVS Forum Special Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 1,147
Mentioned: 206 Post(s)
Tagged: 0 Thread(s)
Quoted: 870 Post(s)
Liked: 1213
Quote:
Originally Posted by madshi View Post
Wasn't sure if you were waiting for a new test build or not, so I made one just to be safe:

http://madshi.net/madVRhdrMeasure83.zip

Same as before, only with the +5% fixed.
Thank you, it wasn't really a problem, but it's less confusing now with the real values.

After removing the cuts from black and high APL changes from the real scenes, 30% seems to be the sweet spot for me:




I didn't test for the less overall misses here, only for the lowest false positives while detecting most of the real scenes (looking also if Metric1 was high enough to compensate).

As the difference is only 1 less false positive compared to 25% and 30%, it would be better to confirm with some more examples...
madshi and Manni01 like this.

Last edited by Neo-XP; 05-19-2019 at 11:49 AM.
Neo-XP is offline  
post #5917 of 8106 Old 05-19-2019, 11:41 AM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
Thank you!

I don't see 30% in the Excel sheet?

So how do we proceed from here? You need/want more testing time before we decide on a percentage value?
Neo-XP likes this.
madshi is online now  
post #5918 of 8106 Old 05-19-2019, 11:59 AM
AVS Forum Special Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 1,147
Mentioned: 206 Post(s)
Tagged: 0 Thread(s)
Quoted: 870 Post(s)
Liked: 1213
Quote:
Originally Posted by madshi View Post
I don't see 30% in the Excel sheet?
Oops, I didn't change all the column titles, it's fixed now.

Quote:
Originally Posted by madshi View Post
So how do we proceed from here? You need/want more testing time before we decide on a percentage value?
I will try to find some more examples to confirm, but it shouldn't be far from 30% in the end, seeing how Metric2 reacts to the percentage change.
madshi likes this.
Neo-XP is offline  
post #5919 of 8106 Old 05-19-2019, 12:05 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
Ok, I'll wait a bit. If you're done with testing, please let me know, then I can create the very last Excel test build...
Manni01 and Fer15 like this.
madshi is online now  
post #5920 of 8106 Old 05-19-2019, 12:21 PM
Senior Member
 
Join Date: Jan 2018
Posts: 394
Mentioned: 100 Post(s)
Tagged: 0 Thread(s)
Quoted: 225 Post(s)
Liked: 346
Neo,

Here's an excel doc containing only the real scene cuts from my test scenes that you don't have @ 35%, 30% and 25% (so if you want, you can copy/paste them)

https://mega.nz/#!w2QhQAKb!zoJbBdi68...m99NpEAXAOFfrQ

(I haven't tested build 83 so I've made a separate excel doc from the build 81 results)

PS. Fantastic Beasts: The Crimes of Grindelwald is a rather high APL change for the scene cut (frame FALL goes from 100 - 473)




Quote:
Originally Posted by madshi View Post
Ok, I'll wait a bit. If you're done with testing, please let me know, then I can create the very last Excel test build...

Music to my ears!
madshi and Manni01 like this.

Last edited by Fer15; 05-19-2019 at 12:28 PM.
Fer15 is offline  
post #5921 of 8106 Old 05-19-2019, 12:28 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
@Fer15 , I assume these are the percentage values from test build 82? (Test build 81 didn't have any percentage values below 35%). If this is not from build 83 then the percent value in the Excel sheet is 5% too low, right? Or did you already correct that? When you selected 35% in test build 81, you actually got 40%...
madshi is online now  
post #5922 of 8106 Old 05-19-2019, 12:36 PM
Senior Member
 
Join Date: Jan 2018
Posts: 394
Mentioned: 100 Post(s)
Tagged: 0 Thread(s)
Quoted: 225 Post(s)
Liked: 346
Quote:
Originally Posted by madshi View Post
@Fer15 , I assume these are the percentage values from test build 82? (Test build 81 didn't have any percentage values below 35%). If this is not from build 83 then the percent value in the Excel sheet is 5% too low, right? Or did you already correct that? When you selected 35% in test build 81, you actually got 40%...

The 35% is from build 81, and I just added the 25% and 30% stats from build 83.


Oh, I thought that only build82 had the mishap when you introduced the 5% increments?



Quote:
Originally Posted by Neo-XP View Post
@madshi are you sure the displayed percentages are correct in madVRhdrMeasure82 ?

35% in madVRhdrMeasure82 gives me higher values than 35% in madVRhdrMeasure81.

30% in madVRhdrMeasure82 gives me the exact same values as 35% in madVRhdrMeasure81.


Yeah, I just checked:


35% with build 83 is giving me the same result as 35% with build 81
madshi likes this.
Fer15 is offline  
post #5923 of 8106 Old 05-19-2019, 12:41 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
Quote:
Originally Posted by Fer15 View Post
The 35% is from build 81, and I just added the 25% and 30% stats from build 83.
Perfect!
madshi is online now  
post #5924 of 8106 Old 05-19-2019, 01:57 PM
AVS Forum Special Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 1,147
Mentioned: 206 Post(s)
Tagged: 0 Thread(s)
Quoted: 870 Post(s)
Liked: 1213
Quote:
Originally Posted by Fer15 View Post
Here's an excel doc containing only the real scene cuts from my test scenes that you don't have @ 35%, 30% and 25% (so if you want, you can copy/paste them)
Thanks, I added them to my sheet, along with some new scenes:




It's all between 30 and 35% (I also quickly tested 40% to be sure not to miss anything).

So, I don't know... maybe testing 33.3% (100/3) to see if it can get the real scenes of 30% and 35%, with the same number of false positives, could help to choose?
madshi, Manni01, tommarra and 2 others like this.
Neo-XP is offline  
post #5925 of 8106 Old 05-19-2019, 02:03 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
Seems like my "randomly" chosen 35% was already pretty good. If you want, I could make a test build with 33.3%, but looking at your test data, I'd be perfectly fine simply keeping 35% and moving on. What do you think?
tommarra likes this.
madshi is online now  
post #5926 of 8106 Old 05-19-2019, 07:49 PM
Newbie
 
armyplace's Avatar
 
Join Date: Nov 2012
Posts: 12
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 12 Post(s)
Liked: 1
Had a quick read of the first few pages of this massive thread.

So is it safe to say that Projectors regardless of whether or not it supports HDR, we set up the projector run in SDR mode and then use MADVR software to apply the HDR levels?

Assumption is your source material has the HDR metadata.

Thanks.
armyplace is offline  
post #5927 of 8106 Old 05-19-2019, 08:16 PM
AVS Forum Special Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 1,147
Mentioned: 206 Post(s)
Tagged: 0 Thread(s)
Quoted: 870 Post(s)
Liked: 1213
Quote:
Originally Posted by madshi View Post
Seems like my "randomly" chosen 35% was already pretty good. If you want, I could make a test build with 33.3%, but looking at your test data, I'd be perfectly fine simply keeping 35% and moving on. What do you think?

I think let's move on
madshi, Manni01, RioBar4U and 1 others like this.
Neo-XP is offline  
post #5928 of 8106 Old 05-19-2019, 09:15 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 497 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6642
Quote:
Originally Posted by armyplace View Post
Had a quick read of the first few pages of this massive thread.

So is it safe to say that Projectors regardless of whether or not it supports HDR, we set up the projector run in SDR mode and then use MADVR software to apply the HDR levels?

Assumption is your source material has the HDR metadata.

Thanks.
Yes, technically any display on earth can use this.

The best part is, it makes any display on earth have the same 'best there is' tone mapping for HDR.

Consistency finally.
SOWK and Iarejake like this.

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #5929 of 8106 Old 05-19-2019, 09:44 PM
Newbie
 
armyplace's Avatar
 
Join Date: Nov 2012
Posts: 12
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 12 Post(s)
Liked: 1
Quote:
Originally Posted by Javs View Post
Yes, technically any display on earth can use this.

The best part is, it makes any display on earth have the same 'best there is' tone mapping for HDR.

Consistency finally.
This is awesome news! What's the minimum requirement in terms of the CPU/GFX card combination for this to work without dropping frames?
armyplace is offline  
post #5930 of 8106 Old 05-19-2019, 09:58 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 497 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6642
Quote:
Originally Posted by armyplace View Post
This is awesome news! What's the minimum requirement in terms of the CPU/GFX card combination for this to work without dropping frames?
No idea sorry.

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #5931 of 8106 Old 05-19-2019, 11:59 PM
Advanced Member
 
tommarra's Avatar
 
Join Date: Aug 2013
Location: USA
Posts: 635
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 641 Post(s)
Liked: 384
Improving Madvr HDR to SDR mapping for projector

Quote:
Originally Posted by armyplace View Post
This is awesome news! What's the minimum requirement in terms of the CPU/GFX card combination for this to work without dropping frames?


On my RTX2070 with almost no enhancements, minor scaling to 4096x2160, low Chrome upscaling, black bar detection, HDR tonemapping I get no dropped frames with render times of 30-38ms for 4K 24fps content.


It also worked on a test RTX2060 - no problem whatsoever just going Lower on some setting and selecting some trade offs. I would say minimum you need is perhaps a 1660 or 1070 to get a good experience. But that’s just my guess based on my current system



Sent from my iPhone using Tapatalk

Gear: Fronts: B&W 803D3 | Center: B&W HTM2D3 | Surrounds: B&W 705S2 | Backs: B&W DM601S3 | Heights: Definitive Techonlogy ProCinema 1000 | Receiver & Amps: Anthem MRX 720 + Emotiva BasX-5 | Projector: JVC NX7 | Source: HTPC with Nvidia 2070, Kodi + External Player: MPC-HC with MadVR (for Blu-ray rips), Nvidia Shield (for streaming content)
tommarra is offline  
post #5932 of 8106 Old 05-20-2019, 02:37 AM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
So here comes the next test build:

http://madshi.net/madVRhdrMeasure84.zip

This will probably be the last Excel sheet style test build. The options are as follows:

1) "resolution": The resolution in which madVR does motion vector search. Resolution "2" has 4 times more pixels than resolution "1". So it probably also costs 4x as much speed. Consequently, I'd prefer to pick resolution "1", which is also what all the recent test builds used. But I'll let you test resolution "2", just to be sure we're not missing a potentially dramatic quality improvement.

2) "search distance": The max supported motion vector distance. In theory higher should be better because it will detect motion better in very high motion scenes. However, supporting larger motion vectors also means a higher chance to detect a false positive motion vector. So this might not be as clear cut as it seems. A big problem with testing is that supporting a large search distance will only help with those few video samples which really have such large motion vectors, while it might harm a little bit in all other video samples. Generally, I tend to prefer going as large as possible here, though. Unless it comes with a significant quality loss.

3) "block size": The area around each pixel which is compared to find a matching motion vector. In theory going large here is "good" because it helps reject false positives better. However, if there's a small object moving around, we might not detect the movement properly if we require too large a neighborhood of pixels to match. So going too large will also hurt in some cases. So we need to find the best compromise.

Both search distance and block size also depend on the resolution. For example, both block size and search distance must be doubled if we double the resolution, to achieve the same effect. So for the bigger resolution, probably we also need to use bigger search distance and block size, which again costs more speed. One more reason to pick the lower resolution, if it's possible without losing too much quality.

FYI, the setting "resolution: 1, search dist: 2, block size: 3" should produce the same results as the previous test build at 35%. So you don't need to retest "resolution: 1, search dist: 2, block size: 3", you can just copy the numbers.

P.S: We already tried testing some of this a while ago, but it was before we have the "threshold" logic in the Excel sheet, and Neo's and Fer's results heavily contradicted each other. So I'm hoping this time around with the increased number of test scenes, and the "threshold" logic in the Excel sheet, we'll get more conclusive results.
SOWK, SamuriHL, markmon1 and 11 others like this.
madshi is online now  
post #5933 of 8106 Old 05-21-2019, 01:37 PM
AVS Forum Special Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 1,147
Mentioned: 206 Post(s)
Tagged: 0 Thread(s)
Quoted: 870 Post(s)
Liked: 1213
@madshi



Doc: https://www.mediafire.com/file/cs9h4..._V16.xlsx/file

Best: resolution 1, search dist 2, block size 2 (if we follow the threshold logic)

PS: maybe it would make sense to test 30% and 35% again to be sure it doesn't change anything now?
madshi and Manni01 like this.

Last edited by Neo-XP; 05-21-2019 at 01:40 PM.
Neo-XP is offline  
post #5934 of 8106 Old 05-21-2019, 02:09 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
Wow, that's 20 instead of 25 misses, and this while consuming less GPU power. What's not to like about that!?

I've compiled new ax files for you for testing 30% and 33%, please copy these over test build 84. Settings dialog will be the same as test build 84. So you can switch between 30%, 33% and 35% by replacing the ax files:

http://madshi.net/madVRhdrMeasure84neo.zip

(I'd suggest to only test 1-2-2.)
SamuriHL and Manni01 like this.
madshi is online now  
post #5935 of 8106 Old 05-21-2019, 02:11 PM
Member
 
VerGreeneyes's Avatar
 
Join Date: May 2013
Posts: 141
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 21 Post(s)
Liked: 17
Quote:
Originally Posted by madshi View Post
Both search distance and block size also depend on the resolution. For example, both block size and search distance must be doubled if we double the resolution, to achieve the same effect. So for the bigger resolution, probably we also need to use bigger search distance and block size, which again costs more speed. One more reason to pick the lower resolution, if it's possible without losing too much quality.
If I understand this correctly, would it make sense to compare say 1-2-3 with 2-4-6? I assume this would quickly get very slow, but just so we aren't comparing apples to oranges (in Neo-XP's most recent comparison, 2-2-2 does look like a strict improvement over 1-1-1 with the exception of Passengers 47773 in real scenes, though it has some different misses in terms of false positives).
VerGreeneyes is offline  
post #5936 of 8106 Old 05-21-2019, 02:17 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
Quote:
Originally Posted by VerGreeneyes View Post
If I understand this correctly, would it make sense to compare say 1-2-3 with 2-4-6? I assume this would quickly get very slow, but just so we aren't comparing apples to oranges (in Neo-XP's most recent comparison, 2-2-2 does look like a strict improvement over 1-1-1 with the exception of Passengers 47773 in real scenes, though it has some different misses in terms of false positives).
It would make sense if "2" were exactly double "1" etc. But that's not the case. It's just 3 different sizes for search distance and block size, while resolution actually quadruples the number of pixels. So simply "doubling" everything isn't possible. Also, I can't really go higher than the max options I'm offering now without completely destroying speed. So "4" or "6" is completely impossible to achieve in decent speed.
madshi is online now  
post #5937 of 8106 Old 05-21-2019, 02:27 PM
Member
 
VerGreeneyes's Avatar
 
Join Date: May 2013
Posts: 141
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 21 Post(s)
Liked: 17
Aah, gotcha. Well, looking just at resolution 1 seems to make sense then, and 1-2-2 does seem to strike a very nice balance between real misses and false positives
VerGreeneyes is offline  
post #5938 of 8106 Old 05-21-2019, 02:31 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
Yeah, I'm happy with the test results. After doing some very little tests myself, I wasn't sure which resolution was better (but hoped for the smaller one for speed reasons), and it looked like the middle block size might be best. And I was already convinced that we should go with the larger search distance, to support as large a motion vector as possible. So the test results basically fulfill all my wishes...
Icaro, Manni01, Javs and 1 others like this.
madshi is online now  
post #5939 of 8106 Old 05-21-2019, 03:31 PM
AVS Forum Special Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 1,147
Mentioned: 206 Post(s)
Tagged: 0 Thread(s)
Quoted: 870 Post(s)
Liked: 1213
Ok I compared 1-2-3 and 1-2-2, with 30%, 33% and 35%.

The closest to [email protected]% is [email protected]%, with the same number of overall misses, but instead of 10 false positives and 10 real scenes misses, it's 12 false positives and 8 real scenes misses. Not worth it.

After adding the missing real scenes from Fer15 sheet (thank you Fer15 for sending them to me), this is the final result:



Final doc: https://www.mediafire.com/file/1iner..._V17.xlsx/file
madshi, Icaro, Manni01 and 2 others like this.
Neo-XP is offline  
post #5940 of 8106 Old 05-21-2019, 03:39 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,856
Mentioned: 579 Post(s)
Tagged: 0 Thread(s)
Quoted: 2757 Post(s)
Liked: 3887
Great, thanks! So we have now oficially completed Metric2 tuning. No more Excel sheet tests.

Tomorrow I'll release a new test build with additional options to combine Metric1 and Metric2 in new ways.
stevenjw, SamuriHL, Icaro and 7 others like this.
madshi is online now  
Sponsored Links
Advertisement
 
Reply Digital Hi-End Projectors - $3,000+ USD MSRP

Tags
dynamic tone mapping , hdr , madvr , sdr , ton mapping

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off