Image Quality and Framebuffer Speculations for WIP/alpha/beta/E3 games *Read the first post*

Thinking of the other PS3 to PS4 ports (and of course their XBox equivalents), I don't doubt it won't be 1080P, at least on PS4, with maybe an upscale or dynamic resolution on the X1.

Was there a single last game port on PS4 that wasn't 1080P? They were even going for 60Hz in a lot of cases, as well. Metro Redux, a game that surely takes advantage of the new hardware, managed 60Hz. As did TLOU. Tomb Raider tried, but didn't reach it. Sleeping Dogs runs pretty badly, though. But all of them at 1080P. So, instead of trying 60Hz, Rockstar went the other route and added a lot of stuff to the game, making it look a lot more "next-gen" than before, plus they added a lot of other things (increased traffic, etc), as well. And Rockstars developers are wizards with GPUs (well, at least on consoles), with what they got out of the old ass last gen hardware, running at a framerate that was... barely okay.

I just hope they didn't go too far. I want rock solid 30hz for GTA5, if I can't get 60Hz.
 
can we get some analysis on the new Uncharted 4 footage?

http://www.gamersyde.com/download_uncharted_4_a_thief_s_end_psx_gameplay_demo-33743_en.html

Here are some screenshots I took, all real time on PS4 (the guy doing the live demo at the playstation experience actually accidentally got killed, although the video I linked doesn't show that)

ibl9kT4H42lr63.png

ixRoyJDYrhAsg.png

i39HVZscT0069.png

ic2L9Cn6GdZ6U.png

ibyfZlUtFewJtZ.png

ifTU7f8TQbyu0.png

iJ5wZlZn2ryuz.png

ibowNZwtlu9NKP.png

ifSQ9nL4QIwNk.png

iYbUTUqC3Otuc.png
iMCw08BCrSKuX.png

iOAS1NSOCr6xf.png
 
Is it me or the colours are quite muted and non-contrasty, and the lighting not as in your face as other games which makes it look a little bit flat? Pre-alpha so it doesn't really matter, I guess.
 
Is it me or the colours are quite muted and non-contrasty, and the lighting not as in your face as other games which makes it look a little bit flat? Pre-alpha so it doesn't really matter, I guess.

The black levels in the video weren't set right, so it will look a bit more desaturated than the final game, but the color palette of that level certainly has a desaturated feel to it as it starts at night, but by the end of the level things start getting brighter as the level ends in daytime. Uncharted games always have a variety of different looking levels so it's a bit early to judge it as everything we've seen so far comes from this same level.
 
The black levels in the video weren't set right, so it will look a bit more desaturated than the final game, but the color palette of that level certainly has a desaturated feel to it as it starts at night, but by the end of the level things start getting brighter as the level ends in daytime. Uncharted games always have a variety of different looking levels so it's a bit early to judge it as everything we've seen so far comes from this same level.
I loved the vibrant colours in the Uncharted games and it would be a shame if ND went for a more muted, more 'realistic' colour palette, going for dark and gritty like everyone else. Who wants bloody reality anyway??:D
Again, way too early to worry about this but still, we can discuss endlessly like we usually do.
 
Is it me or the colours are quite muted and non-contrasty, and the lighting not as in your face as other games which makes it look a little bit flat? Pre-alpha so it doesn't really matter, I guess.
The RGB levels aren't matched in the video (at least the gamersyde video) so it looks washed out. You can tell by inspecting it with a histogram.
I'm gonna try to re-encode the video and preserve as much detail as possible, but it's hard to do while keeping the size reasonable because the video was already encoded/compressed. My encode will probably be a bit larger than the gamersyde video (my guess is ~2GB).
 
Last edited:
The RGB levels aren't matched in the video (at least the gamersyde video) so it looks washed out. You can tell by inspecting it with a histogram.
I'm gonna try to re-encode the video and preserve as much detail as possible, but it's hard to do while keeping the size reasonable because the video was already encoded/compressed. My encode will probably be a bit larger than the gamersyde video (my guess is ~2GB).
No need, this is a colour corrected video and the best one around.

https://mega.co.nz/#!NBcA3BiK!AB4PDO0ni2ZvTPDpjXmsQRu2Ttaw-KuvAZFMtlDQdwo
 
No need, this is a colour corrected video and the best one around.

https://mega.co.nz/#!NBcA3BiK!AB4PDO0ni2ZvTPDpjXmsQRu2Ttaw-KuvAZFMtlDQdwo
Was that re-encoded? 3.81GB is excessive IMO. At most you'd need 2.5GB for a 'transparent' encode. Even 2GB looks very close.

With that said, that version is probably slightly better than gameryde's re-encoded video, but almost double the size.
gamersyde have now a colour corrected video as well for download...
Damn, wish I knew they were working on this. I haven't actually started encoding the video, but I spent a while figuring out how to fix it, and then doing a few test encodes to determine how much bitrate was needed. Oh well, saves me some bandwidth.
 
Last edited:
Damn, wish I knew they were working on this. I haven't actually started encoding the video, but I spent a while figuring out how to fix it, and then doing a few test encodes to determine how much bitrate was needed. Oh well, saves me some bandwidth.

for future reference, if you're using x264, just use CRF encoding and a quantizer of 10 or lower for pretty much visually lossless video. 0 is actually lossless, but for a 1080p video it would be really huge and most of the benefits under quantizer 10 aren't usually noticeable by the human eye.
 
I have a decent amount of experience with x264, not so much with avisynth scripts. Most of the time was spent figuring out how to convert the RGB range to fix the black levels.

And in regards to quality/testing, I know I could just plug in some settings and know that it's pretty much identical to the original. But I like to do testing to find the most efficient settings (ie deblocking strength, aq strength, psy-rd/trellis, optimal amount of b-frames etc.,), and I like to know for sure how much bitrate and/or CRF value is actually necessary for a 'transparent' encode.

The original gamersyde video was already encoded/compressed, so it left zero room for compression, which is why all of the re-encoded videos are larger than the original. I tried to retain quality while keeping the size reasonable.
 
Last edited:
Well the same CRF is always going to give you the same quality, but yeah you can optimize the file size if you test changing those other settings I suppose. Also just setting the x264 preset to very slow though would probably do a good job optimizing the size if that's what you're worried about. While it would take a lot longer to encode, it would still take less time than figuring out a bitrate to encode manually, and produce probably comparable results.
 
I agree for viewing purposes, but for analyzing the footage for resolution, AA, frame rate, etc, 18 may be a bit high
Ya, the video is HQ so it's good enough for judging IQ. I didn't want to cause any artifacts from additional compression. CRF 18 would introduce a bunch of artifacts on close inspection (blocking, edge shimmer etc.), but would probably be good enough for the average joe.

I downloaded Gamersyde's encode and we ended up using around the same CRF. I used 15.2 and they used 15, however our other settings differed which affect the bitrate of the encode, so mine was actually slightly bigger. Comparing them side by side with AvsPmod, they look pretty much identical. So rest assured that gamersyde's re-encode is very good and basically transparent to the original.
 
Last edited:
Back
Top