He is somewhat inaccurate and he likes to throw tech words around a lot to oversell...this DF article is much better.
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Digital Foundry - The Order 1886 Tech Analysis
- Thread starter CozMick
- Start date
He is somewhat inaccurate and he likes to throw tech words around a lot to oversell...this DF article is much better.
OdysseusVA
Banned
Pretty much this. It's tough making it through his Uncharted 4 vid.
Nitpicking? Are You even serious?
Yes, good that there are few textures with PoM, but they are definitely rare, thats why could have been missed.
And yes, IQ is only slightly better than 900p.
And DoF was 1/4 res in the past.
1:05 - http://www.gamersyde.com/hqstream_the_order_1886_e3_trailer-32302_en.html
I don't even know if you're serious with the bolded statements alone. I've been through the IQ part many times but it seems you still decide to stick with hyperbole, so I'll just ignore it. But I have no idea how you can judge the DoF in the final build from a demo a year ago. It's like judging God Of War 3 from the E3 2009 demo and saying "Yeah, it doesn't look so good". Did you even take the time to judge the final build? I don't think so. You were insisting it was quarter res all the time and when I posted a pic with DoF, you vanished.
Dictator93
Member
DoF is the most expensive component of the post-processing pipeline. It generally stays in the same resolution, i dont even remember example of a game where resolution for DoF was upped from preview to final game on consoles.
What a pity that we dont have screenshots without post effects, so You would understand how destroyed clarity of image is.
---
It was friday, i got back home after 3am ...
---
Generally agree, especially his analysis of U4 demo was total bullshit, but since then he got more knowledgeable.
His videos definitely get better and better.
It doesn't matter if an effect is extremely expensive. You can't jump to conclusion and immediately assume it won't improve. And BTW, I'm not denying this game looks a tad blurry but calling it destroyed the IQ is hyperbolic. The heavy post processing fits the atmosphere of the game very well and many agree with me.
Terbinator
Member
What is he inaccurate on or "overselling" though? It all seems correct to me!
Do note that DoF remains in those shots. Motion blur is disabled along with other minor differences.
Dictator93
Member
The images are not showing up for me... eiher in quote form or otherwise. odd!
Going back to the U4 video (I have yet to watch any others). It had lot to do with term usage and or inferring without looking for direct evidence of something which points out a technique.
http://images.eurogamer.net/2013/articles//a/1/7/3/8/2/0/0/xps4_000.bmp.jpg/EG11/resize/1920x-1
http://images.eurogamer.net/2013/articles//a/1/7/3/8/2/0/0/xps4_001.bmp.jpg/EG11/resize/1920x-1
Images not showing, but i know which one You are talking about. They still have grain filter and CA.
--
I dont jump to conclusion. Its called educated guess and sure i can be wrong sometimes.
I agree it fits the game and You take 'destroyed' word to strong. And it also depends what standards do You have for IQ.
Like i said before, i'll take slight blur over aliasing any day of the week while playing.
While it's not as good as the DF article..
He noticed the lips sticking together..
I'm still stunned by this.
1.21Gigawatts
Banned
But this The Order video video seems to be pretty correct.
Only thing that weird is that he didn't mention the lack of proper AF, unless I missed that.
I never saw his U4 video but watched this and thought it was detailed and full of good tech information. Nothing seemed wrong to me, infact the soft shadows I mentioned from my play-through he picked up and was not in the DF article at all.
I'm inclined to agree. IQ is more than sharp edges. I much prefer a lower incidence of aliasing than needing sharp edges to everything. The heavy post processing in this game really settles characters and objects into the world.
It was mentioned but the wording seemed as they had no idea what sort of technique the game was using :/
It also helps the game to look "more CGI", so I really don't mind the heavy PP.
Dick Justice
Banned
Fanboy insecurity.
LOL, you can't use the word "destroyed" liberally and blame others for "taking it too strong". "A bit too blurry for my taste" is a lot more reasonable and others won't take it the wrong way.
I mean, because of the post processing sometimes You cant even see textures properly on screenshots. Thats a bit more than 'a bit blurry'. The post-processing is literally destroying textures, in the same way FXAA or upscaling does. I just find this word fitting the process.
Dictator93
Member
I watched the video and he definitely has gotten much better (still resorts to adjectives and phraseology which makes me question his motives though), but there is some handwavey stuff with the decrtiption of SSR in compariosn to how the Order does reflections. As in, to paint one as being technlogical inferior to the otehr when they in reality, both have pretty huge trade offs.
There are a number of places, like in tesla's lab or in any place with ibls and more indirect specular, where SSR is much much better. Likewise, I think harping on the cubemaps as being impressive is somewhat odd. They are higher resolution than in other games... but they are just cubemaps.
EDIT: BTW dark10x, you mention the specular reflections from lights in the review at some point. Did you check and see if they are point lights or actual lights with area such as in KZSF?
Ok, this is not true. You can still clearly see the details of the textures. Sure, they may not look very sharp, but the textures still look very good.
Sure You can, but pretty similar to upscaled games from 900p
Bench and wall textures here http://i.imgur.com/sPztmfC.png
900p games
http://abload.de/img/screenshot-original4p6jmk.png
http://abload.de/img/screenshot-original399k7v.png
http://i.imgur.com/yIg3Qic.jpg
Omotesando
Member
Wow that part of the video where it switches to Crysis. The Order is truelly a new benchmark. It looked like a true generation leap.
Dictator93
Member
Crysis 3 is 2 years old and is pre-pbr among other things. This game better have a more physically accurate renderer or soemthing is wrong.
Omotesando
Member
Yea Crysis is one game. But any game really?
That textures in the first shot look amazing to me and better than the others, so let's just agree to disagree.
Dictator93
Member
I really do not think a generational leap is a great way to describe it. Crysis on low vs. Very High looks generationally different.
The Order game is doing a lot of things other games do too (and other games even do the same things better), it is just that this game has a meticulous attention to detail and a consistent artistic vision.
Generational? Nah. Consistently great looking? Yeah.
In their presentation they said that they dont have area lights, because they are quite problematic in forward+. And going by speculars it seems to be a case.
--
Yep, consistency is amazing in The Order. They were aiming at pretty much similar blend art and technology like God of War 3 and it payed off.
Personally, i still get over beach landing scene from Ryse, this is the craziest scene in real-time rendering for me.
Those nice sharp lines are nicely aliased. I'll bet the effect looks even worse in motion.
Hmm, i thought lamps had volumetric lighting, it doesnt seems to be a case on the zeppelin playthrough.
Are they volumetric in other game sections or they just fake this volumetric effect globally?
----
Are they volumetric in other game sections or they just fake this volumetric effect globally?
----
Wait what? Are we really going into delusional zone?
Consistently better looking than other games. Like I said before something about this game screams CGI. It looks astonishing. To see so much potential in this wonderful setting wasted by the pathetic gameplay is soo saddening
No surprise you can add details and post-fx when developing for a sub iPad (heck even sub FullHD) resolution with 30(!) fps. Instead of appreciating this it should be taken with a pinch of salt.
sotheresachance
Banned
Does the game run smooth? I was watching dansgaming stream yesterday and there were parts where it wasn't fluid at all. Was it the stream or the game?
Its the stream.
http://www.eurogamer.net/articles/digitalfoundry-2015-the-order-1886-tech-analysis
This is one very impressive game.
What a stupid comment. The iPad runs at 2048 by 1536 pixels these days. Most high end gaming PC's would have trouble running demanding games at full settings at 30fps+ with resolutions that high, let alone a freaking console.
Dictator93
Member
That is a rather uninformed comment.
Not really, posting from my own relatively high end gaming PC where I've spent the past few days messing with the settings/NVIDIA effects and DSR in Far Cry 4. Just an oversimplification. 2048x1536 is not some "low" resolution that a game should be scoffed at for being below, especially a console game.
thunder_snail
Member
So, RAD has pretty much set the bar for the most impressive graphics for now.
Who'd be the first one to surpass it? Crytek with Crysis 4? Naughty Dog with Uncharted 4? Quantic Dreams with whatever they are working on? Sega with Shenmue 3?
Who'd be the first one to surpass it? Crytek with Crysis 4? Naughty Dog with Uncharted 4? Quantic Dreams with whatever they are working on? Sega with Shenmue 3?
Some nuance is required here. that resolution probably isn't too taxing for a relatively high end PC (even mid -end) to play iPad quality games. Anything else (like the Order), then yes, that resolution is quite high, especially when considering the +.
Insane Metal
Member
This, so much this. I can't stand playing the game because the FOV is absolutely horrible. It's waaaay to low.
Dictator93
Member
That is what I dont get about this. All benchmarks I have ever looked at show "slightly" higher than 1080p resolutions being rather fine to run @ 30fps.
http://www.hardocp.com/article/2015/01/28/msi_geforce_gtx_960_gaming_high_resolution_review#.VOjOai5Mf-s
GTX960 isnt even high end...
You're not going to get those options from console games. If the black bars got turned off in this game, then you would see IQ tank.
Seeing as IQ is mostly what this game has going for it, you're not going to get those options.
Yeah. Just looking at the average fps, then fine. But the fps often drops below that for some period of time (I don't know what the horizontal axis describes).
These days I kind of prefer to use the minimum fps as a better metric of consistency. It isn't a good experience if the frame rate consistently tanks (doubly true if operating at 30fps max).
There is far too much fanboying and not enough TECH going on in this thread. I'm about to change this.
Every since the first DF thread on The Order, there have been a lot of arguments going on about the relative "sharpness" of its image quality. Since I like discussing technical aspects with a solid technical foundation, I thought it would be very useful to have some sort of objective metric for "sharpness".
Before going into what I came up with, I'd like to remark that sharpness is not, in and of itself, a virtue. It's only one factor in IQ, and a very sharp picture can also suffer from terrible spatial and temporal aliasing.
That out of the way, what is sharpness really? I'd argue that it makes sense to define the "sharpness" of an image by the amount of high-frequency detail it shows. Or, to put it more accurately, the amount of high-frequency detail in an image is indicative of its sharpness, as with low sharpness you also can't get any high-frequency detail.
Based on this idea, I've written a very simple Matlab script which illustrates the Fourier transform of images. You can read these like this, somewhat simplified: low-frequency (large-scale) image artifacts will cause higher values towards the center, and the closer you get to the image borders the higher frequencies you see.
As an initial test, I wanted to see if this type of illustration gives any important information in a scenario which is rather clear. Therefore, I chose two equivalent screenshots from DF's comparison of the Ryse PC and XB1 versions. One of those renders at 1080p, the other at 900p while upsampling, and they are almost exactly equivalent otherwise.
Ryse XB1
Ryse PC
As you can see, the absolute magnitude of the FFT is significantly higher further out towards the edges for the PC version, indicating its higher level of fine-grained detail visible due to the higher rendering resolution. This is a good result for my methodology, I'd say.
However, Ryse on PC rendering at native resolution (without downsampling) is still not a game anyone would call particularly "sharp". It uses lots of postprocessing and rather heavy AA. Therefore, as a really really sharp (almost "too sharp") baseline, I tested the script on an old Oblivion screenshot, I had, with no postprocessing AA (only MSAA and sharp downsampling).
Downsampled Oblivion with MSAA
Well, that's that.
So now, how does The Order fare? The following is one of 3 screenshots I tried, which looked quite similar in their overall impression in the frequency magnitude space -- this is the most representative of the 3.
The Order 1886
Based on this, my conclusion is that The Order 1886 is sharper than 900p Ryse on XB1, slightly less sharp than Ryse at 1080p on PC, and a lot less sharp than rendering without postprocessing / post-AA (duh!).
Every since the first DF thread on The Order, there have been a lot of arguments going on about the relative "sharpness" of its image quality. Since I like discussing technical aspects with a solid technical foundation, I thought it would be very useful to have some sort of objective metric for "sharpness".
Before going into what I came up with, I'd like to remark that sharpness is not, in and of itself, a virtue. It's only one factor in IQ, and a very sharp picture can also suffer from terrible spatial and temporal aliasing.
That out of the way, what is sharpness really? I'd argue that it makes sense to define the "sharpness" of an image by the amount of high-frequency detail it shows. Or, to put it more accurately, the amount of high-frequency detail in an image is indicative of its sharpness, as with low sharpness you also can't get any high-frequency detail.
Based on this idea, I've written a very simple Matlab script which illustrates the Fourier transform of images. You can read these like this, somewhat simplified: low-frequency (large-scale) image artifacts will cause higher values towards the center, and the closer you get to the image borders the higher frequencies you see.
As an initial test, I wanted to see if this type of illustration gives any important information in a scenario which is rather clear. Therefore, I chose two equivalent screenshots from DF's comparison of the Ryse PC and XB1 versions. One of those renders at 1080p, the other at 900p while upsampling, and they are almost exactly equivalent otherwise.
Ryse XB1
Ryse PC
As you can see, the absolute magnitude of the FFT is significantly higher further out towards the edges for the PC version, indicating its higher level of fine-grained detail visible due to the higher rendering resolution. This is a good result for my methodology, I'd say.
However, Ryse on PC rendering at native resolution (without downsampling) is still not a game anyone would call particularly "sharp". It uses lots of postprocessing and rather heavy AA. Therefore, as a really really sharp (almost "too sharp") baseline, I tested the script on an old Oblivion screenshot, I had, with no postprocessing AA (only MSAA and sharp downsampling).
Downsampled Oblivion with MSAA
Well, that's that.
So now, how does The Order fare? The following is one of 3 screenshots I tried, which looked quite similar in their overall impression in the frequency magnitude space -- this is the most representative of the 3.
The Order 1886
Based on this, my conclusion is that The Order 1886 is sharper than 900p Ryse on XB1, slightly less sharp than Ryse at 1080p on PC, and a lot less sharp than rendering without postprocessing / post-AA (duh!).
Dictator93
Member
They have the 30fps line in there. These benches are at an even higher res than the ipad res.
Edit:
Gonna have to read through this and think! Thanks for the obviously high-content post.
Edit 2: you mind linking that modded oblivion shot?
Well, those benchmarks kind of support my point. They show 30-40ish performance at higher than 1080p for BF4 with 2xAA, but they don't test 4xAA at higher than 1080p, only at 1080p. Ditto for Far Cry 4, where they only use the cheap SMAA for all resolutions. Performance at 2560x1440 would surely drop in both cases at 4xAA in BF4 or using TXAA or MSAA in Far Cry 4, etc.
I did say full settings in my post, and I'd consider full settings to be higher than 2xAA or SMAA. Like I said, it was an oversimplification for the sake a of a quick remark.