The Hobbit Trilogy 4K UHD Extended & Theatrical (2012-2014)

[levcore]

Originally Posted by wildphantom

I concur and have said it many times. The Hobbit is a cracking trilogy. So what, it isn't as good as Lord of the Rings. What is? As a fantasy epic it's probably got as close to the Rings trilogy in terms of quality as any film/s in the last 20 years - and boy have other studios tried!

Exactly!

[PissedOffPeoN]

I just love the way it flows into the lord of the rings when marathon ed all the way .

[weberrygt]

I totally agree with recent discussion. As is often the case, the most discussion, whether positive or negative, is generated from something that is truly remarkable. LOTR and Hobbit were both fantastic trilogies, and together were a magical journey. I can't wait to see them both in 4k!

[philthehip]

Originally Posted by goathead1

Does anyone think that the Lord of the Rings would have suffered the same criticism as The Hobbit if The Hobbit had been released first?

Im not talking about the CGI aspect because there are people I've met that simply don't like The Hobbit since it came out so many years after the Lord of The Rings movies.

That is impossible to say. If they had made The Hobbit first it would never had have to adhere or fit in with what was shown first. Then The Lord of the Rings would have differed too.

Going off what you ask and expanding a little. I felt that the makers missed a trick with The Hobbit, they should have revised the plot to suit the bigger story they were trying to incorporate by shifting the White Council to a time after The Hobbit rather than forcing the stories together in a way Tolkien himself considered wrong and abandoned (1960 Hobbit). That shift would have meant all the Sauron stuff could have been left out of The Hobbit (2 movie) with a third dedicated to the White Council, Hunt for Gollum and the build up to The Lord of the Rings. It would have also meant that fans would get more Gollum which would have been met with enthusiasm.

[Deciazulado]

The Hobbit The Motion Picture Trilogy 4K Blu-ray

[slumcat]

Originally Posted by Vangeli

I’m not saying just multiply the frames by five then divide by four, I’m questioning if blending to 240fps then cutting down to 60fps is better than going from 48-60 using current frame blending techniques. Most of the frames used in the final 60fps render would be newly calculated frames using the difference data between two original frames and there would be zero pull-down judder.

Not sure I can explain it better tbh. Might have to open up some programs later.

I think I understand what you are saying and if I think about it, your way (48-240-60) WOULD create better quality than simply going 48-60.

Though I am guessing one measure of the success of interpolation might be how many original frames are still maintained in the presentation.

In a 48-240-60 conversion, 20% would be original frames and 80% would be interpolated frames.

In a 48-60 conversion, it could potentially be the reverse, 80% would be original frames and 20% would be interpolated frames.

That is if we take the assumption to be true that interpolated frames would necessarily be inferior in quality to original frames.

[Pyoko]

It wouldn't make any difference to go to 240 first instead of directly to 60, both would result in the same cadence, which is one real frame per second, the rest being generated ones.

If you were to only generate the missing 12 frames every second and tried sticking them in there among the real 48 ones, you'd get some kind of weird pulldown judder similar to just repeating a frame every fifth frame. I've never tried what happens if you do it this way, but I can imagine it might still be noticeable because although the frame would be interpolated from the surrounding frames instead of just repeated, it would still mean an unnatural break in the speed/flow of the video.

I'm working on some high-speed Phantom footage right now, might be an interesting test to try out.

[weberrygt]

Originally Posted by weberrygt

[Show spoiler]I totally agree with recent discussion. As is often the case, the most discussion, whether positive or negative, is generated from something that is truly remarkable. LOTR and Hobbit were both fantastic trilogies, and together were a magical journey. I can't wait to see them both in 4k!

How did this post of mine end up in both the Hobbit and LOTR threads? I only remember posting this in the LOTR post, but here I see it in the Hobbit thread. My profile says both were posted at the same time. Go figure!

Anyway - my post still applies. I can't wait to see both in 4k!

[slumcat]

Originally Posted by Pyoko

It wouldn't make any difference to go to 240 first instead of directly to 60, both would result in the same cadence, which is one real frame per second, the rest being generated ones.

If you were to only generate the missing 12 frames every second and tried sticking them in there among the real 48 ones, you'd get some kind of weird pulldown judder similar to just repeating a frame every fifth frame. I've never tried what happens if you do it this way, but I can imagine it might still be noticeable because although the frame would be interpolated from the surrounding frames instead of just repeated, it would still mean an unnatural break in the speed/flow of the video.

I'm working on some high-speed Phantom footage right now, might be an interesting test to try out.

I agree that with the 48-> 60, you might essentially end up with 1 real frame per second and 59 generated ones - that is if you interpolate EVERYTHIHG, which I agree would result in smoother motion than just interpolating 12 frames.

But 48 to 240 to 60 will let you keep 12 real frames and 48 generated frames per second.

So if higher number of real frames per second is a desirable outcome, then 48 to 240 to 60 is better.

[Deathbymonkeys]

Originally Posted by Alister_M

I'm sure they're fine on 3D BD, I just happened to get put off cinematic 3D projections after DOS being botched. Doesn't help that I saw it in a smaller, older theater which had obviously been poorly retrofitted for 3D, rather than the larger one in which I saw every other 3D presentation prior to that.

Oh man I LOVED the HFR but I saw it in good theaters. I also watch the 3d at home and it's amazing. Sorry you had a bad experience but maybe try again if you ever get the chance. The best looking 3d movies ive ever seen.

[Pyoko]

Originally Posted by slumcat

I agree that with the 48-> 60, you might essentially end up with 1 real frame per second and 59 generated ones - that is if you interpolate EVERYTHIHG, which I agree would result in smoother motion than just interpolating 12 frames.

But 48 to 240 to 60 will let you keep 12 real frames and 48 generated frames per second.

So if higher number of real frames per second is a desirable outcome, then 48 to 240 to 60 is better.

The interpolation/retiming functions work with floating point numbers which don't always round so nicely so it's possible none of the frames except maybe the first and last ones in any given shot will technically be completely real anyway, even though it's probably going to make an insignificant difference in most cases whether a frame gets designated as having to display frame 14 exactly or frame 14.025. But after the decimation of the 240fps roundtrip surely you're going to end up with the same subframe numbers? If you have any perfectly real frames/even frame numbers in there, e.g. every fifth frame or whatever, then they're going to be in the straight 48->60fps version as well because the 48/60 factor evens out on the same frames as 48/240?

[Alister_M]

Originally Posted by Deathbymonkeys

Oh man I LOVED the HFR but I saw it in good theaters. I also watch the 3d at home and it's amazing. Sorry you had a bad experience but maybe try again if you ever get the chance. The best looking 3d movies ive ever seen.

PQ wise they were great. First one was fine after that initial muckup, but I just don't like how it mimics that "soap opera" effect. Maybe if films had always been done at a higher frame-rate I'd be able to accept HFR more readily, but the motion blur is just too intrinsically embedded in my mind as part of the cinematic experience.

[Zombie Dude]

Originally Posted by weberrygt

How did this post of mine end up in both the Hobbit and LOTR threads? I only remember posting this in the LOTR post, but here I see it in the Hobbit thread. My profile says both were posted at the same time. Go figure!

Anyway - my post still applies. I can't wait to see both in 4k!

Someones gone and split the threads up because god forbid we discuss the Hobbit in a LOTR's thread

At least these two trilogies work together to create one cohesive adventure story that spans generations. I'd understand more with Star Wars where the new trilogy feels "woke" and doesn't flow well with what was established prior.

[slumcat]

Originally Posted by Pyoko

The interpolation/retiming functions work with floating point numbers which don't always round so nicely so it's possible none of the frames except maybe the first and last ones in any given shot will technically be completely real anyway, even though it's probably going to make an insignificant difference in most cases whether a frame gets designated as having to display frame 14 exactly or frame 14.025. But after the decimation of the 240fps roundtrip surely you're going to end up with the same subframe numbers? If you have any perfectly real frames/even frame numbers in there, e.g. every fifth frame or whatever, then they're going to be in the straight 48->60fps version as well because the 48/60 factor evens out on the same frames as 48/240?

The way I see it is this way.

Say the 1st 10 seconds of footage in 48 FPS are represented this way

1R
2R
3R
4R
5R
6R
7R
8R

with R indicating Real frame and the preceding number representing the frame number.

Say you interpolated 48 fps up to 240 fps, you could represent it this way

1R - 1G1 - 1G2 - 1G3 - 1G4 -
2R - 2G1 - 2G2 - 2G3 - 2G4 -
3R - 3G1 - 3G2 - 3G3 - 3G4 -
4R - 4G1 - 4G2 - 4G3 - 4G4 -
5R - 5G1 - 5G2 - 5G3 - 5G4 -
6R - 6G1 - 6G2 - 6G3 - 6G4 -
7R - 7G1 - 7G2 - 7G3 - 7G4 -
8R - 8G1 - 8G2 - 8G3 - 8G4

with G representing the generated frame, the preceding number representing the preceding real frame and the following number representing the position of the generated frame after the preceding real frame.

If you interpolated this 240 fps down to 60, you could represent it this way

1R - 1G1 - 1G2 - 1G3 - 1G4 -
2R - 2G1 - 2G2 - 2G3 - 2G4 -
3R - 3G1 - 3G2 - 3G3 - 3G4 -
4R - 4G1 - 4G2 - 4G3 - 4G4 -
5R - 5G1 - 5G2 - 5G3 - 5G4 -
6R - 6G1 - 6G2 - 6G3 - 6G4 -
7R - 7G1 - 7G2 - 7G3 - 7G4 -
8R - 8G1 - 8G2 - 8G3 - 8G4

with underlined and red-colored frames representing 60 fps.

So in theory, you should end up with every 5th frame being the real frame in 48 -> 240 -> 60. So that is at least 20% being real frames which is better than straight 48 -> 60 which would only be 1 real frame and 59 generated frames.

[Pyoko]

Originally Posted by slumcat

So in theory, you should end up with every 5th frame being the real frame in 48 -> 240 -> 60. So that is at least 20% being real frames which is better than straight 48 -> 60 which would only be 1 real frame and 59 generated frames.

It depends on how the algorithm works and how the time-stretching/interpolation is defined, for example if you set the first and last frames in a shot to be real frames, and interpolate everything linearly inbetween, then chances are low anything else in there will be a perfect original frame. However I was mistaken in my initial reasoning, if you only "anchor" the first frame in a shot, define that as the first real frame, you would achieve the exact same thing going from 48 directly to 60.

Going from 48fps to 60fps is a time-remapping function with a factor of 48/60 = 0.8. So for every frame in the 60fps result sequence, we only advance 0.8 frames from the 48fps source sequence.

The function then becomes:

Y = 1+0.8*(X-1)

Where X is the 60fps output frame and Y is the 48fps source frame (we have our 60fps sequence and for each frame we want to know which frame we should display from the 48fps source sequence). A non-even frame like 7.4 means a generated frame, interpolated from frames 7 and 8, but slightly closer in time to the former. If we step through the frame numbers we get a result like this:

X Y
1 1.0
2 1.8
3 2.6
4 3.4
5 4.2
6 5.0
7 5.8
8 6.6
9 7.4
10 8.2
11 9.0
...
56 45.0
57 45.8
58 46.6
59 47.4
60 48.2
61 49.0

As you can see you get the exact same result where every fifth frame is an even number, i.e. an original, non-generated frame.

[slumcat]

Originally Posted by Pyoko

It depends on how the algorithm works and how the time-stretching/interpolation is defined, for example if you set the first and last frames in a shot to be real frames, and interpolate everything linearly inbetween, then chances are low anything else in there will be a perfect original frame. However I was mistaken in my initial reasoning, if you only "anchor" the first frame in a shot, define that as the first real frame, you would achieve the exact same thing going from 48 directly to 60.

Going from 48fps to 60fps is a time-remapping function with a factor of 48/60 = 0.8. So for every frame in the 60fps result sequence, we only advance 0.8 frames from the 48fps source sequence.

The function then becomes:

Y = 1+0.8*(X-1)

Where X is the 60fps output frame and Y is the 48fps source frame (we have our 60fps sequence and for each frame we want to know which frame we should display from the 48fps source sequence). A non-even frame like 7.4 means a generated frame, interpolated from frames 7 and 8, but slightly closer in time to the former. If we step through the frame numbers we get a result like this:

X Y
1 1.0
2 1.8
3 2.6
4 3.4
5 4.2
6 5.0
7 5.8
8 6.6
9 7.4
10 8.2
11 9.0
...
56 45.0
57 45.8
58 46.6
59 47.4
60 48.2
61 49.0

As you can see you get the exact same result where every fifth frame is an even number, i.e. an original, non-generated frame.

Your algorithm and mine are identical.

So I guess we can conclude that going from 48 to 240 to 60 should essentially get the same result as going from 48 to 60 with an equal proportion of real frames preserved.

[Deciazulado]

1/4th of a sec of a Hobbit 48fps to 60fps example:



1 2 3 4 5 6 7 8 9 10 11 12 frames @48fps
x5 =
11111 22222 33333 44444 55555 66666 77777 88888 99999 1010101010 1111111111 1212121212 repeat frames @240fps
/4
1111 1222 2233 3334 4444 5555 5666 6677 7778 8888 9999 9101010 10101111 11111112 12121212
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 frames @60fps

100 25/75 50/50 75/25 100 100 25/75 50/50 75/25 100 100 25/75 50/50 75/25 100 frame blending % @60fps

=
6 100% original frames (40%)
6 slightly blended (blurred) 75/25 frames (40%)
3 fully blended (blurred) 50/50 frames (20%)


vs the @ 24fps 1/4th second 12 reduced to 6 50/50 frames, in other words, up to 6 frames of fully blended (blurred) 50/50 frames to prevent strobing.

60fps vs 24fps would only have 1/5 of the fully blurred frames of 24fps and 40% of original frames, with the 40% sligthly blended frames, lets say 40% x 0.25blend = 10% blurriness + 20% fully blurred = 30% decreased motion res vs 100% decreased motion res


The way the eye motion retention works I don't think there would be much juddering.

Do any of you see lots of intolerable juddering when the BBC interpolates their 50 programs into 60 discs? Have you compared a motion interpolated 60 Who or Planetoid Terra disc directly to their original 50 counterpart?

And I would think motion res on the 60fps UHD would end, not 2x better than on a 24fps discs, but about 1.5x better, not counting any improvement that a dedicated AI analyzing the start stop points and throttling down or up the blending percentage of each pixel in the frame.


c'est la vie en bleu

[hoejunter]

Me reading this thread:

[Jumpman]

Originally Posted by hoejunter

Me reading this thread:

Exactly. 🤣

[vicious_boy]

Originally Posted by hoejunter

Me reading this thread:

I know, right