On the use of Photoshop to enhance images from Second Life
Posted by Burgess
Myg’s post on Photoshop and Second Life images sparked a good conversation and raised some important issues. I started to comment, but then got so long-winded I figured it would be better to lay my thoughts out in another post.
Let me begin by saying that I am not a programmer, nor a graphics expert of any kind, either in 3D rendering or Photoshop. I’m just a voracious reader with a lot of curiosity. I don’t claim to understand the technical details of the Second Life graphics engine, nor am I going to explain them. However, I am going to set out my understanding of why it is the way it is. If I’m wrong, please correct me.
No matter how good the graphics platform, card, or skill of the photographer, Photoshop will always be able to “improve” an image. This will be true no matter how much better the graphics engine of Second Life becomes (and let’s hope it gets better quickly and continues to do so).
For the user/viewer, there will always be a difference between interacting in a “3D†graphic environment with motion and viewing a two-dimensional still image of that environment. Photoshop is the best tool (so far) to enhance and create ever more stunning *still* images. Second Life, like other MMOs, is about a sense of place or a feeling of being “inside” that the user gets through interaction. Viewing 2D still images is a representation of that, but is something different itself–a visual art form all its own. Even when those 2D photos are of a “3D” environment, be it SL, WoW, or first life.
Complaints about graphics quality in Second Life are common and understandable. Linden Lab made a big tradeoff in putting the ability for users to generate content ahead of graphics quality.
In a virtual environment where users can create the objects and textures, the processing power required is so great that it ultimately limits the graphics display. When you add in the vastly varying skill levels of second life residents, most being at the novice level, you get a lot of “poor†graphics.
In a virtual space where the parent company controls nearly every aspect of the design, such as WoW or any of the high-end console games, graphics quality can be maintained. Not only because the designers are professionals who spend countless hours refining images, but because the processing power of the servers, software, and users’ computers is not compromised by also having to keep track of user generated content that is malleable and constantly changing.
However, in those environments, you don’t have the ability to build a house, texture it, retexture it, tear it down and build another, all the while changing your avatar’s outfit so that you can go from your construction foreman clothes to your entertaining-in-your-new-pad-cocktail-hour look. All this simultaneously for 40,000+ residents online, plus keeping track of all the content of the other half-million who have contributed who aren’t concurrently logged in. If SL didn’t need to keep track of all these constantly changeable parameters, a lot of processing power could be dedicated to simply presenting rich graphics. At present, Second Life’s less-than-perfect graphics capabilities include actual hardware limits, not just coding.
Some (many?) images may be so striking, or have a certain rawness that is desirable, that using Photoshop will not help. Almost all of the pics we use here at M is for Myg are nearly untouched. Typically we:
1. adjust levels
2. adjust brightness
3. adjust contrast
4. resize
Most of the pics are decent. We work hardest at composition of the shots inworld. And we try to have good costumes and sets, which means good textures. The difference between a good texture and a crappy texture is the difference between a good photograph and shit.
For example a good (and funny) photo Myg took of herself:
And a crappy photo I took without much thought:
Both of these were used on the blog. Myg took care with the setting (place, textures, costume) and with the position of the camera. The second was taken by me, on the fly, without much effort (too great a distance). And the textures are crappy. Dull, and boring. Neither of these photos was enhanced beyond the aforementioned adjustments.
Now take a look at the original of the photo Myg posted in the previous post about Photoshop and Second Life. Please click through to examine the large sizes of the images.
Because we’re writing a graphic blog serial, eventually Myg and I would like to have the skill and time to have images like the following enhanced version accompany all our story posts:
While the most obvious change is that the background was swapped out for a “real†background, a close examination shows that significant work was done to the image to improve it. The background has been blurred to give the impression of depth we are used to in first life photos, an effect that is a result of the focal length setting on real camera but is not present to the same degree in our eyesight (hopefully) nor is it in a 3D virtual environment.
Adjustments have been made to the color balance, brightness, contrast and hue, bringing the fore- and mid-ground into alignment with the “evening†lighting.
A lot of work has been done to the edges to smooth out roughness in the original. This work is one of the most effective uses of Photoshop for SL images. Vint has a good video tutorial on how to use liquefy to smooth edges here [link] and a text tutorial on using gaussian blur to add that sense of depth that you see in the shot of Myg on the roof of her loft (above) and in the shot of Lisa and Vanny below. One of my all-time SL faves. Check out the image below that is a RL photo of a print of the SL shot:

The end result is one that looks significantly better than the original, yet still looks like a snapshot of Second Life—meaning that it hasn’t been transformed into an art image beyond its original state. Such transformations are also wonderful applications of Photoshop—really using the tool to its utmost to create original art. Here are some examples (they’re on others’ Flickr accounts, please click through):
Stella Errans - Pokinoï
oxoc - thirdeyeblackwall
nathaliebee



















hehe such a long post but it was very interesting to read, i honestly dont know whats such the big deal with with photoshoping pictures. if it makes the picture more pleasing why not. plus its a skill that not everyone has so it makes the pictures even mroe valuable. i would be totally happy if your friend photoshoped one of my pics, i would even consider it an honor that one of my pics inspired him enough to take the time to make it better.
i also i wanted to ask…*taps her first fingers together while looking down blushing* i really like the mesh top in the last pic but i cant find it anywhere…so hummm…where can i find it? ^^
I know, sometimes when I get going I really can’t shut up. And what are blogs for? Glad you made it to the end.
I asked Myg, she thinks it comes from Nyte n Day.
thanks ^^ i’ll check it out right now ^^
Hi everyone,
Coming form the post you’re referring, I’d like to add something:
In SL, avatars are meshes (complicated 3D objects) whereas everything else you can build is made of prims (very low polygons objects). I understand the need for processing power. But graphics processing is made client side not server side, meaning that the quality of the graphics depends on the user machine not LL (that’s why you can adjust your graphics preferences depending on your hardware capabilities). With all the great graphic cards out there able to run very beautiful games like Half life 2, why would SL be limited to the current graphics ?
There are never more than 60 people at the same time on one sim (or less it’ll lag very badly). How many multi-player games have great graphics and allow more than 50+ people to play with each other (Unreal/Quake-like games for example).
From the server side, their is a need for storage of the 3D objects information and good bandwidth for delivering that information (plus other considerations like general environment considerations : whether sims are linked, sun/night levels, wind, etc.).
Now there is one problem though (that you pointed out in your post) : SLifers can build their own world at anytime. Due to this feature, it can “seem” hard to obtain good graphics without a lot of processing power as objects’ shape/texture/position are changing all the time.
Well this problem can be divided in smaller problems and get addressed easily.
First : most sims are no-build/no-scripting except for the sim owner (or owning group). Thus the only moving objects are scripted ones and avatars (btw, I think scripted objects are simulated client side : the server just need to know the orginial position/shape of the object.). And as I said earlier, avatars are already computed easily.
Now when the owner of the sim is building objects, he won’t build a thousand objects at the same time (i mean really at the same time), it’s one object after another.
And no more than 15,000 objects per sim. Why not allow meshes objects then ? Meshes would be uploaded to the server and sent back to the clients whenever needed.
We don’t need to manipulate the meshes in-world : they can be created and textured with tools like Maya or 3Ds max and uploaded when we’re done with it (just like sculpted prims).
For the sims where building is allowed to anyone (like sandboxes) that’s a bit more difficult : the problem here is not computing power, it’s more broadband. If anyone is able to build complicated 3D objects at the same time, you need to send them to the server and back to the users everytime objects get modified.
In this case there are solutions to optimise broadband :
First, 3D meshes would not be modified in-world, only their position and content (just like sculpties). Only this info is needed by the server then (cause the mesh info has been uploaded when rezzed).
Next, if LL releases the server-side source code of SL, servers could be managed and installed by third-parties. This would help non-american SLifers get a better SL experience. And we (i’am european), would not “cry” when a server is lagging if we know that it’s an australian one (for instance).
Next : if there are only 50+ people max on one sim, why not link them together in a P2P way ? People from the same geographical area would see a significant increase in speed and graphics quality.
As for linked-sims, they could be installed on the same server. Indeed, LL does not guarantee that linked sims relies on the same server. This causes some more lag and other network issues.
Well, sorry for the long post but I think it needed some clarification.
Here’s another thoughtful take on it:
Rodion’s myspace blogpost “Why do SL graphics suck?”