Larian Studios Forums Beyond Divinity (English) General RR Quality Concerns.

..... .............

RR 768 VR ...............LH 600 VR

Pictures speak for themselves. <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" />

<img src="/ubbthreads/images/graemlins/riftrunner.gif" alt="" /> 800x600


Lionheart 800x600



<img src="/ubbthreads/images/graemlins/wave.gif" alt="" />

I can't because I don't use such a high resolution.

And I wouldn't because at such a high resolution Role-Playing in itself seems riduculous to me.

I mean - I play the role of a person that has the height of a Fairy in other games ? How am I supposted to do good role-playing if I almost cannot see the the char whose role I'm playing ?

Bingo. <img src="/ubbthreads/images/graemlins/up.gif" alt="" />
That is why I am suggesting the modifications that are necessary to make playing an RPG at 1600x1200 becomes a pleasure rather than being a ridiculous idea.
My technical calculations produced a value saying that within our current technological state we need the avatar to be 200 pixels high for super-deluxe quality, or 175 pixels high for an acceptable and fabulous quality, while a lower bound limit is 120 pixels for an acceptable and enjoyable quality but without anything special about it.

Given that the market insists on considering the 800x600 to be still the standard around (which I strongly doubt), a character would be one third of the screen tall and consequently there would be no space for any interesting scene but it is perfect for zooming in and object manipulation. With 1024x 768 the character becomes a quarter of the screen rather than a third and a perfect resolution for a melee battle. With 1200x1024 the avatar becomes 1/5 of the screen’s height and the player may enjoy the high quality from close distance if the battle field is not to be seen as a whole. 1600x 1200 is the inevitable destination of computer based games and the reason behind that is the simultaneous quantity of information per screen and the avatar is at perfect quality with a height of 1/6 screen while being informed of all the immediate environment.

Fixing the details of the avatar is only a benchmark of course but it does not work that way.
Let us say that the majority of PC users have 1024x 768 screens and graphic cards (which is a fact).
Zooming in on the avatar should bring the lower spec relation where the avatar is 1/3 of the screen and here the avatar should be 256 pixels high in relation to the resolution being used. Normally you would zoom to normal with ratio = ¼ and an avatar being 192 pixels high with full details. Now zooming out to 1/5 and 1/6 should bring the avatar’s height to 154 and 128 pixels respectively. This specification is concerned with the dominating resolution and the minimal number of pixels that carry information. If there was someone out there with an 800x600 screen zooming out to a 100 pixel should be acceptable at such low resolution, but why do we have to suffer all of us as if we all have that system and at a maximum zoom out? You see, the 200 pixels were the benchmark of all resolutions to calculate the ratios only. We concluded 1/3, ¼, 1/5 and 1/6 to be the four dominating ratios of the avatar to the screen height. Now with one resolution being your favourite option And let us recall the dominant 1024x 768 we still apply all those ratios for zooming in and out. This means that your maximum zoom in shows your avatar 256 pixels high and your maximum zoom out shows the avatar as 128 pixels high. The currently implemented height in RR screenshots is between 100 and 110 pixels, which really does not make a big size difference but it makes all the difference with details.

I really hope that Marian should give it a try and experiment with avatars being at least 128 pixels high and then you can see the difference I talk about.

Cheers.

The avatars are always in motion, even when they stand still, they breath and move. So no way.



That's why i doubt that we support that resolution, at least officially. <img src="/ubbthreads/images/graemlins/winkwink.gif" alt="" />



The avatar is always in the highest resolution. The textures will only get shown with lower resolution in higher screen size. When zooming in you will see much more details then.



The engine will play as much frames as it can tough the game(animation) speed is 24 frames. One model (Hero or NPC) has around 1000 Polygons. Afaik there is no cutting of details whatsoever. Only the textures get midmapped so Details might get blurry.



I can only answer things from the graphic artist point of view. However i must admit that sometimes i just don't get what he is actually talking about. <img src="/ubbthreads/images/graemlins/shame.gif" alt="" />
If just feels not like we did it.



Now thats impossible. <img src="/ubbthreads/images/graemlins/winkwink.gif" alt="" />

That much I have figured out by comparing the standards of T3D games on the market.
Well, I guess RR and LH could stay each on a nipple. <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" /> <img src="/ubbthreads/images/graemlins/evilgrin1.gif" alt="" />

Good luck Marian.
I shall buy RR anyway because I know that the interface is far much better than that of LH.
For me the playability takes the ultimate importance.

Cheers.

BTW; It is probably too late to change anything in RR so let us focus on DII when you are through with RR.

Until the game is out there is always room for changes. This topic you Guys brought up for instance made us compare the different looks of the avatars in <img src="/ubbthreads/images/graemlins/div.gif" alt="" /> and <img src="/ubbthreads/images/graemlins/riftrunner.gif" alt="" /> too. Well, there is work in progress now... <img src="/ubbthreads/images/graemlins/winkwink.gif" alt="" />

i just can't wait till the game is out <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" />
glad to see you pay close attention to "us" <img src="/ubbthreads/images/graemlins/up.gif" alt="" />

well thanks 2 DAD for clearing matter up in terms of res. still i feel RRR's avatar packs more punch(when it's zoomed in) though when zoomed out looks grainy(due 2 mipmapping as what marian said).

sad for DAD, no 1600 X 1200 for now at least. for those who play on computers with 15" - 17" monitors, 800 X 600 is ok. 1024 X 768 may look a little small unless scaling is proper or zooming feature is in 2 help. also sad for DAD, the idea of having static avatar on high quality while animated avatar on relatively lower quality isn't going 2 be realised as well since marian says the avatar on stance will still be animated.

marian, i hope u can do something about the grainy effect.

if i'm not mistaken(i have 2 re-install NWN or some other T3D games), certain games have options 2 install lower res avatar or higher res avatar. if zoom factor is fixed(1/4X, 1/2X, 2X, 4X), what about making different sets of avatar 2 match the zooms so u won't have 2 magnify/zoom out the avatar & have something like grainy effects showing?

just a flash of my mind bulb. i think it blew. <img src="/ubbthreads/images/graemlins/idea.gif" alt="" />

marian, when u said there's room for improvement, u don't mean overhaul of graphics, right?

I apologize for making assumptions that lead to writing birds flying over heads. <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" />
Here is an attempt to be as simple as I can.
The T3D model of an avatar is created by the help of some utility but whatever that utility is we end up with a mathematical definition.
Let us say that the avatar is defined in a space with a coordinate origin at the centre of its pelvic bones like a marionette. Also let us assume that each avatar’s surface is described by 1000 ± 100 polygons all in all.
Each polygon is defined relative to that origin by three coordinates and each coordinate is defined by three numbers, thus each polygon is defined by a total of nine numbers. I hope this is clear enough.
Now each polygon is a two sided lamina, and to identify the outer surface a number called the normal-to-the-surface-direction is a solid angle relative to the axis originating from the point of origin. Now bear with me.
Each animation script, for example walk, run etcetera must have a corresponding list of frames. Each frame list must include 10000 numbers because each polygon of the 1000 requires 9 coordinates plus one surface normal direction. Now given the eye relative position to the world and the heading of the avatar in motion, with simple arithmetic a decision could be made regarding the visible polygons in each frame. The number of visible polygons in each frame must be less than half the total number of polygons per frame. That is why by testing 1000 numbers only we can decide the less than 4500 numbers we need to read per frame. The associated textures are list index relative and they are called by polygon index number, then by some mathematical equation for tilt we can compress the bitmap into the 2D area that represents the visible polygon.

Now read very carefully here.

If that area that represents a polygon was too small we cannot avoid graininess and bad quality.
So, how we can find out if such an area was too small or not, is by calculating the minimal vertical number of pixels the count of which corresponds to visual information of identifiable features.

Now let me demonstrate the super deluxe model first.
I shall assume zooming out to the maximum and displaying the avatar as small as possible.
Now let us examine a vertical dissection passing into eyes’ centre and nose.
A single polygon for the head top (I hope there is no less than one because zero would let us see his brains).
A single polygon for the forehead (a white space of skin colour)
A single polygon for eyebrows (a colour of hair)
A single polygon for the eyes
A single polygon for the nose
A single polygon for the moustache/ upper lip
A single polygon for the lower lip
A single polygon for the chin.
This scheme is definitely out of proportion but let us just assume that this is acceptable as the least resolution.
Therefore the head needs a minimum of 16 vertical pixels so that we may call it a head. <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" />
Dwarves and imps shall solve a lot a problems here so take them not as any benchmark. <img src="/ubbthreads/images/graemlins/smile.gif" alt="" />
Let us assume that our avatar is a gorgeous babe the head of which is one part.
Japanese people have shorter legs and longer abdomens, while African people have longer legs and shorter abdomens. Now let us take the international average which states that a head is 1/6 of the total body height excluding children of course. 16 x 6 = 96, therefore 96 pixels are absolutely needed for the worst case of a super deluxe model.
To add one pair of pixels for the head is to add 12 pixels as an increment for the whole body. Therefore the next better model is 108 pixels high with a nose two pixels long. <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" />
For having a head top that looks any better than a box lid we need two pixels.
Therefore the least acceptable model must be 120 pixels high for an adult human avatar.
Add to those, 8 pixels for anti-aliasing to blend the avatar softly with the background, similar to diffractive effects of light in nature. This brings our standard avatar’s height in pixels to a binary practical number.

This means that my original polygons must be distributed proportionally to fill a vertical scale, which is a multiple of 128 if we shall use integer arithmetic or just 128 if we shall use floating point arithmetic.
The final outcome is displaying every feature rather than discarding intermediate zeros.

This has nothing to do with zooming in or out or screen resolution or scale. It is the basic information required for the minimal full detailed adult human avatar. Putting the same information on 256 pixels doubles the polygonal displayed height and the details of texture but no new polygons are being added here.
This means that zooming in enhances the textures only but does not add polygonal details. In fact zooming in may add previously discarded polygons, which might seem as if we added new polygons but we did not do that.

My whole calculation was focussed on avoiding the discarding of polygons on zooming out.
If the adult human avatar was displayed in less than 128 pixels high there is no safeguard against loosing polygons and consequently some details, which we perceive as blurriness and/ or graininess.

If this still flies over heads I have no idea what I could say, but I shall try if I was asked to do so. <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" />
<img src="/ubbthreads/images/graemlins/wave.gif" alt="" />

[color:"yellow"]
Addendum: I have edited the post to correct some numerical glitches and to add those pictures for comparison.
[/color]


...Height = 64 PIX...Bad

...Height = 96 PIX...Threshold

...Height = 128 PIX...Good

Naturally I would prefer this height. <img src="/ubbthreads/images/graemlins/winkwink.gif" alt="" />

...175 PIX ...Excellent
<img src="/ubbthreads/images/graemlins/kissyou.gif" alt="" />

I can't agree more DAD. The larger and more detailed that female warrior is the better <img src="/ubbthreads/images/graemlins/badsmile2.gif" alt="" /> <img src="/ubbthreads/images/graemlins/badsmile2.gif" alt="" />

i made a little modification, what do you think of this 185 pixel version? <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" /> <img src="/ubbthreads/images/graemlins/evilgrin1.gif" alt="" />

175 pixels 185 pixels

i will call it the Morbo expansion pack <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" />

@bean: now those are Quality Concerns. The virtue of photoshop. this takes me back to that thread where this warrior would have higher charm skills <img src="/ubbthreads/images/graemlins/badsmile2.gif" alt="" /> <img src="/ubbthreads/images/graemlins/badsmile2.gif" alt="" />

ontopic: I just thrust the good judgement of Larian studio's to make it great. And like my ethic teacher says discussion is good if its a good discussion <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" /> <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" />

Pity about the other measurements: 28-20-30 pixels. Looks a little anorexic to me.

@ Bean
<img src="/ubbthreads/images/graemlins/evilgrin1.gif" alt="" />

LOL; this is horrible.
I do not endorse the F cup size. <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" />

so what size do you support.(and yes F cup is a bit large)

Now that the hero is in 3D. Wouldn't it be easy for users to create theire own items. you just need a skin to put on the polygons. think about it
the armour of morbo
Thick green exosceleton
gloves with claws
helmet of morbo: +100 inteligence <img src="/ubbthreads/images/graemlins/tongue.gif" alt="" /> <img src="/ubbthreads/images/graemlins/tongue.gif" alt="" />
...

or edit your hero I would like to play like a char that looks like my ava.
Maybe an editor would be nice for the grafic designers among us <img src="/ubbthreads/images/graemlins/winkwink.gif" alt="" /> <img src="/ubbthreads/images/graemlins/winkwink.gif" alt="" />

@DAD: theire is no alternative motive behind this post my motivations are pure and good so stop looking <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" /> <img src="/ubbthreads/images/graemlins/biggrin.gif" alt="" />

I endorse the C and the D only. <img src="/ubbthreads/images/graemlins/evilgrin1.gif" alt="" />

Also I am not going to look; here, you see, I am turning around and facing the wall, yes, that wall that have that mirror on it.
<img src="/ubbthreads/images/graemlins/puppyeyes.gif" alt="" />
I am not looking.

Char editor could be like: NeverWinterNights + Baldur's Gate + Dungeon Siege, for the new 3d models.

Principal: You can choose the race, and make a "different race", miscelaneous.

Hey Bean your idea of charm skills isn't so bad.

So when you take your helmet off.... <img src="/ubbthreads/images/graemlins/badsmile2.gif" alt="" />
Are you allowed to wear it in school?
<img src="/ubbthreads/images/graemlins/think.gif" alt="" /> <img src="/ubbthreads/images/graemlins/badsmile2.gif" alt="" />