7 New Year Resolutions for Archviz Artists

Title.jpg

It's late January, but it's never to late to take some good resolutions for the year to come 

1. Referencing

Be open to everything

Whether you're working for a quite broad panel of clients or a specific niche, it is important to keep your mind open to the multitude of references out there. Photography, painting, copywriting, calligraphy, furniture design, fashion, urban design, collage, carpentry, literature, etc. 

There are several ways to get your daily dose of references : pinterest, instagram, feedly or dedicated newsletters from curators. 

Just try to take at least half an hour each day to browse through diverse contents and save what inspire you to your own library for later use.

Image credits : random stuff from inspiration folder. scenography, atelier particulier, mad men series, cogan painting, bloodline, desvignes, etc.

Don't think you're going to reinvent the wheel

Everything's already been done before, or sort of.

That photo you took and thought was original, was painted a century ago by some expressionist painter. That fantasy land you just drew was described by some writer fifty years ago, that composition you thought was original, looks like a daguerreotype from the early 1840's, etc.

That's why referencing is so important. Gather as much information on what's already been done in an area, be aware of the many stones that have already been turned, and confront all this material with your own experience is the best way to achieve something original. Your job is not necessarily to create something unique every time, but to propose the best possible solution to your client's brief.

But, as Paul Arden puts it

"Steal from anywhere that resonates with inspiration or fuels your imagination. Devour films, music, books, paintings, poems, photographs, conversations, dreams, trees, architecture, street signs, clouds, light and shadows. Select only things to steal from that speak directly to your soul. If you do this, your work will be authentic".

    Tidy your bedr... library

    Whether it's for postproduction, modelling or texturing, it is important to tidy at least a bit the libraries you're working with on a daily basis. 

    To be honest, I use a quite simple tree view but it still saves me quite a bit of time. 

    Textures are divided by types of materials (wood, metal, concrete) or types of maps (alpha channel, dirtmaps, etc)

    Cutouts are divided by several tags although I tend to keep them quite broad (mainly to avoid scrolling through people with the wrong kind of clothing, or the wrong view point).

    3D assets are divided by level of details (from furniture layout to types of screws) or size and types.

    Although it might take a bit of time to organize all this at first, you can be sure you'll save heaps of time in the long run.

    Snapshot of the collection panel in Adobe Bridge CS6 and asset manager Connecter

    Make the most out of each project

    At the end of each project comes a quite important phase : cleaning up your desk, whether real or virtual.

    There are plenty of things that you should do at the end of each projects in order to save time on the next ones, for instance :

    • Add all the materials you've created to your own library (that wood texture you've been finetuning for hours)
    • Add all the assets you've created to your own library (now you have a nice ready-made living room layout)
    • Gather all the new references you've used into your inspiration folder, take photograph of your sketches (get new ideas for non-commissioned projects)
    • Make additionnal images of the project you've been working on if you've found an interesting point of view you'd like to dig in order to grow your portfolio. Ask your client for permission if you want to use it on your website (careful with ongoing NDA)
    • Add your freshly finished images to your commissionned-work portfolio
    • Backup your project properly (cloud and on-site) in case you ever need it again

     

    Snapshot of the "add asset" pop-up window in 3ds max from Connecter

    2. Modelling

    Modelling is a big part of making your final image look good and believable. Whether you use ready-made assets or model your own, you have to be aware of the quality of the assets you use in your image and make sure they show enough details. On an opposite note, be aware not to overdetail things that are too far in the background of your image since they will add up in vertices count and render time while not adding anything to the image. You might actually end up with noisier looking asset.

    Chamfer the hell out of your model

    If you look closer to any edge in real life, you'll notice they will never be razor sharp. That's why chamfering all your edges in your model will add instantly a good deal of realism. Done with the boxy look with super sharp shadows transition on the corner of a wall, using chamfer will give you a nice and smooth transition from light to shadow and vice versa

    As for any kind of small details, be aware not to overdo it. You don't want to end up with round-looking corners everywhere. The chamfer we're talking about are about 1 to 2 millimeters. 

    Simple corner modeled in a boxy way and then chamfered using a chamfer modifier in 3ds max

    Use detailed window frames assets

    Another aspect that is often overlooked but that definitely makes a huge difference in an image is the window frame. Whether you're dealing with exterior or interior renderings, using poorly modelled window frame will always end up with weird looking results

    If you don't have time, invest in some good looking assets.If you have time, model a detailed series of window frames and store them properly for reuse in your next projects.

    Window frame, from boxy modelling to more detailed version

    Think like a constructor

    Nowadays, most of the visualizer have a background in architecture, construction or industrial design. This comes in handy when you have to make believable models. For instance, knowing how structure works in real life will help you in modelling believable structural elements. Plus, if you end up with the kind of client that don't have any detailed drawings to provide, you'll be happy to have a proper background and be able to propose working solutions for them so that your final model don't look unresolved.

    A simple wooden structure quickly modeled above, and more detailed below. Image below is the bare minimum you can afford. 

    Use a self-explanatory layer structure

    There are many ways to organize a model, but none of them is good if you don't rely at least a bit on layers. Don't forget to organize your model in different self-explanatory layers so that you can easily isolate furniture, select windows, work on street furniture, or refine your topography in just a couple of clicks.

    Use instances over copies, as much as possible

    Easy tips to follow, that is activated by default anyway. To better optimize your scene, use instances instead of copies. 

    This becomes even more paramount when you're dealing with GPU rendering where Vram (ie : polycounts and textures) can quickly become an issue.

    If you're well organized, try breaking up your model in several xref and combine them in a final file.

    3. Texturing

    Texturing is another big part of a basic workflow to achieve convincing visuals. There are a couple of simple tips to rely on to make sure that you won't fall into the common pitfalls of badly textured visualization

    Use composite and blend material to avoid tiling

    Sometimes, even with the best seamless texture, you mind end up with visible tiling. In order to avoid this odd looking effect (which can sometimes happen in real life, but it is so rare we tend to avoid it), you can simply composite several textures using blending masks and different uvw mapping. 

    Below is an excerpt of Bertrand Benoit's asphalt material for instance.

    Compositing layers and blend materials.jpg

    Snapshot of slate material view for asphalt material by Bertrand Benoit on Nakagin scene. 

    Double check your map scaling

    Another quite visible error in rendering is the wrong scaling on objects. Whether it's a timber structure with giant veins or a leather fabric with unrealisticly large grain, always double check your material mapping scale and make sure it looks realistic. 

    See how the image above has a distorted and out of scale concrete module. The grain of the sofa is too big as well. On the contrary, the image below respects the ratio of the original concrete texture and boasts a more natural grain on the leather.

    Randomize UVW mapping on systemic object

    Identical UVW mapping can also be quite eye-catching and reduce the believability of your image. The most common area you'll see it on is with shelving systems where each shelf has the same exact uvw mapping and map or floor boards. It can also happen on kitchen sets where all the doors of the shelves look identical.

    In order to avoid that, don't forget to randomize UVW mapping (rotate, offset, scale, or blend with another texture depending on the type of material) on your objects. 

    Image above shows repetition of the wood texture on each vertical part while the image below has its uvw mapping randomized.

    Add imperfections to your materials

    Something you quickly notice on objects of our daily life is that there is no such thing as a perfectly clean and smooth surface. There will always be slight imperfection whether it is on the roughness, the reflectivity, the glossiness or the flatness of a surface. 

    These imperfections are quite easy to simulate by adding dirtmaps in your material nodes. 

    There a three main things to keep in mind when dealing with dirtmaps :

    • Scale them properly (don't add a finger print dirt map with 6cm wide thumb print)
    • Avoid visible tiling (try to use a different uvw mapping than your main diffuse texture)
    • Be subtle. Your material will always look better if the variations you bring with dirtmaps are subtle.

    Last but not least, be aware that adding these extra layers of complexity to a material will of course add render time, but it can also create a lot of noise if you're making complex textures for an object that will be too far in the background

    Images from Adán Martin tutorial on realistic copper, available on youtube.

    Use simple and realistic glass material as a base

    Glass material is definitely the one you'll see in 99% of the projects you'll work on. So you'd better get it spot on quickly.

    Even though it is a quite basic material, there are a couple of tweaks you can add to your material in order to make it look a little bit more interesting. 

    Add a tiny bit of bump map so that the surface is not perfectly flat, as well as add a small bulge on the edge of the glass panel. To do so just use a noise map and a simple border hand-made map in your bump slot.

    The next trick is to make slight variations of the same material using different phases of the noise map so that the unevenness of the surfaces are different from one glass panel to another.

    Depending on the distance, you can then start to add dirtmaps on your panels so that they don't look to clean. On the other hand what I sometimes do is use a falloff map in the reflection slot to better control the transition between full reflection and full refraction.

    Image credits : The Boundary, Horoma

    4. Composing

    Use the rule of thirds as a rule of thumb

    Composition is by far the most important part of an image. You can have the best material and lighting, if your image is unbalanced in terms of composition, it will just suck in the end. 

    So before going all crazy in odd composition, you should master the rule of thirds which at least give you the certainty to have a balanced image.

    Once you start to understand how this all works, you can start playing around with it. But not before.

    All these images were composed following (more or less strictly) the rule of thirds

    Create an interesting foreground to anchor your viewer

    Once you've settled on a composition that you like, take some time to work on your foreground. Since this is, well, the foreground, it means this is something that is going to need some special attention to details because it is so close from the camera. 

    This is where referencing comes in handy to get quick ideas of what you could put in the foreground to add that extra touch of details and believability to your image. 

    The other thing you can do is to simply postprocess your foreground in photoshop and blend it nicely into your 3D. 

    But whatever workflow you choose, your foreground needs extra caution.

    First image has no foreground to help you project yourself in the space. Second one has many more details that helps in building the believability of the image.

    Create a believable background to anchor your scene

    The next aspect you should keep an eye on is your background, specifically avoiding to end up with the visible black horizon you would get from your HDRI. 

    You can choose to do everything in 3D or postprocess it, but be careful to have a believable horizon. Concretely, you should know there are rare places in the world where you will get a perfectly flat horizon. There will always be topography variations, hedges, trees, buildings, and other elements like that to make your background more interesting than just a flat line.

    Last but not least, dealing with hiding your horizon in 3D will help in getting believable reflection and not having to rework all of them in postproduction. There's nothing worse than having a visible black ground reflected in your windows. 

    Before and after postproduction. Simple tree lines were added but it really helps in creating that additional depth you're looking for for a convincing background

    Define each plane and what happens in it

    Now that you have a convincing foreground and background, let's take care of the rest. Basically, the idea is to give depth to your image, which means aiming at creating several planes between the foreground and the background, and implement details that make sense for each of these planes.

    Concretely if we have a building, this would mean :

    • Detailing what happens in front of it (sidewalk, plazza, etc)
    • Detailing what happens inside of it
    • Animating each area differently according to the program specification
    • Detailing what happens behind the building
    • And so on

    Keep in mind that animation doesn't necessarily mean adding people. This can be a simple curtain, plants, a basketball, street furniture, birds, etc. 

    If you look at these images you'll see that each plane has its particularity, animation or detail. This builds up the complexity of the image although the composition itself is very simple.

    Double check your verticals, and your horizontals when doing frontal views

    There's nothing worst than a slight offset in verticals or horizontals. Literally nothing more cringy.

    99% of the time, it'll make sense to have your verticals vertical, simply because it looks better this way. So make sure that your verticals are corrected either directly in 3d or later in postproduction.

    The only reasons you wouldn't want your verticals to be vertical is if you're doing an image to emphasize the size and slenderness of a tower (i.e. being on the ground and looking upward to the top of a tower) or to emphasize the depth of an atrium (i.e. standing on a balustrade looking downard). 

    Regarding the horizontals, well, they only need to be horizontal if you're perfectly facing your project.

    Keep in mind that the worst is to be slightly off. Because just like in architecture concepts, in-betweens lack strength. Either do a proper frontal view, or just don't.

    Image above will just drive you crazy while the second one looks more balanced and relaxing

    5. Lighting

    Try out multiple skies

    The sky in an image is of great importance. It can dramatically change the mood of an image, and if it isn't chosen correctly it can also ruin an image. 

    Even if you choose a plausible sky, depending on the kind of image you're dealing with, certain types of sky will work and others won't.

    For example an extremely busy sky (lots of clouds) might not work with an already busy image (many people, and things happening, complex project shape, etc.). That's why you would rather use them with more minimalistic image with simple composition so that the sky can actually take on more importance.

    Be careful with what people call "dramatic" skies, because the outcome can indeed be dramatic. There's a fine balance to find between the emphasis on architecture, the emphasis on animation (people, cars, etc) and the emphasis  sky. A dramatic sky will definitely have a huge impact in the balance of the image.

    Beside that, be careful to avoid the usual errors we can see on many visualizations : 

    • Sun direction in the sky doesn't match the sun direction in the rendering (this mostly happen in interior rendering where you can see the backplate cast shadows don't match you rendering's).
    • Sun intensity in the sky doesn't match the sun intensity in the rendering (overcast sky with razorsharp shadows).
    • Sky hue doesn't match the general hue of the rendering (sunset backdrop with middle of the day 3D lighting). 

    The first sky works, the following don't really. 

    Use light to balance your composition

    Oftentime people think that composition is only a matter of how you place your camera in relation to your project. The thing is, lighting is a huge player in the composition area since this is the discrepancy between dark and bright areas that is going to dictate how you look at an image.

    Therefore, playing with the orientation of the sun (while keeping it plausible) is an important step in finetuning your composition.

    Take the example below for instance, you'll see the same crop of a corner of a project with two different lighting. One gives a flat look while the other creates depth and contrast while retaining the same framing. 

    As a rule of thumb, I suggest to never light both visible facades of a building except if you have a perfectly sound explanation (there's generally no such explanation). Because otherwise you'll end up with a flat looking building and no pleasing contrast in your image.

    Use light to balance your colours

    It is complicated to achieve a balanced and interesting image with monochromatic tones. This is why we tend to always work with a palette of about 3 main hues in an image.

    The good thing that people tend to forget, is that an object can give two different hues and saturation depending on if it's in the light or in the shadow. This means that by actually using the light to balance your composition, you're also creating a more interesting complexity in terms of hues. 

    Using the same example as before, we can see that our bricks take a completely different hue in the shadow, which gives an interesting turn to our image that we didn't have in the fully lit version.

    Use HDRI skies

    Well, this one is not a "must-follow" since you can achieve already good results with a simple Sun and Sky system, but know that you'll always get more interesting shadows and light dynamics with an HDRI file.

    Above is the vray default sky, pretty boring. Below is a snapshot of Peter Guthrie 1224 Clear Sky, a bit more interesting even for a simple blue sky.

    Use smart artificial lighting

    Artificial lighting is a complex topic. Just know that most of the time, in order to create interesting lighting in an image, you cannot only rely on actual spotlights. Softboxes are quite easy to set up, far more flexible than in real life, and will help in getting interesting lighting in your image. Just take time to play around with invisible light planes in your image in order to create emphasis on specific objects. Just like I wrote a little bit earlier, light and shadow plays a part in composition, and softboxes can help you in balancing your composition and emphasizing specific objects over others.

    6. Populating

    Don't make transparent trees or cutouts

    Most of the time, this request will come from your client and not from you. Either way, you should know there are several workarounds in order to avoid relying on unrealistic results while still showing interesting foliage. 

    First if the project is actually supposed to be seen through the trees, special care should be brought to the chosen variety for the vegetation. Proper landscaping will make sure that the size, shape and density of trees and shrubs will make it possible to see the building. Therefore your work will only be to choose your models accordingly.

    An alternative is to render your image in a colder season so that leaves take less space in the image.

    Another alternative, a bit more time consuming, is to carefuly compose your trees so that it only hides bits and pieces of the project while keeping the facade legible.

    Don't pack your images

    It's true a rendering will feel more lively if there are actually people in it. But it is true only to a certain extent. Always keep in mind the actual space you're depicting and make sure the amount of people you're inserting correlate with the program. 

    Whether your driveway is pure architecture genius or not, there will never be 20 people waiting in front of your garage. Same goes with balconies. Sunny days or not, you will never see all your building neighbours go out at the very same time on their balcony. Never.  

    So if you've reached the maximum amount of people you think makes sense for an image, and if your client still ask you to put new people, take time to explain properly why this is a bad idea and how it negatively impacts the project in the end. 

    I'm all for emphasizing the life in the building over the building itself, but there's a fine balance to find that client tend to overlook (one way or the other actually).

    Check the scale of all your cutouts

    Scaling issue is not an issue with 3D people (except if you don't check their scale in your model of course) but can become a nightmare in complicated scenes with 2D cutouts.

    What you usually want to do is drop some dummies or billboards in your 3d scenes in order to get the right scale in various areas of your image and then scale your cutout accordingly.

    Also, if you're dealing with a camera at eye-level, make sure that your cutout align more or less around the horizon. 

    Don't mix puff jacket and swimming suit

    A typical mistake we can see in populated images is the mismatch between the weather and the cutouts, and, worse, between the cutouts among themselves.

    Consistency is the key in populating your image. If people are having a drink inside a place, they probably won't have their jacket on. If the rain is pouring outside, a guy in a tshirt will probably wait it out in a covered area rather than walk carelessly outside.

    Don't forget to foster interaction

    Last but not least : interaction!

    When you populate an image, make sure that people interact with the building and interact with each others. There's nothing weirder than a bunch of people standing in a place not interacting in any way, looking away.

    Show that the place you are depicting is a place where you can socialize. 

    Coming back to the density problem mentioned earlier. Your image will feel much more lively if you put 3 people talking to each other and someone walking by rather than 10 people each going their way in different parts of the image.

     

    You can find out more tips and tricks about cutout integration in our dedicated post here : how to integrate people seamlessly

    7. Postprocessing

    Use an efficient hierarchy layer structure

    Before jumping to the fun of postprocessing, one should have a strong and clear hierarchy layer structure in order to not get completely lost in his file, especially if you have to come back to it later.

    Use non destructive workflow

    We all, once in our life, had to deal with a poorly managed photoshop file where everything was permanently cropped, erased, desaturated, etc. 

    In order to keep maximum flexibility in your work, always try to use non destructive tools

    • Use adjustment layers instead of adjusting the layer.
    • Use layer masks instead of erasing a layer
    • Use smart objects when dealing with objects you're going to resize several times instead of losing quality after several resizing on a rasterized layer
    • Use dodge and burn tool on a 50% Grey layer set to Overlay blending mode instead of working directly on your image

    These are simple habits to take that will save you a lot of time in the long run, as well as give more flexibility to potential users who would try to work on your file.

    Use vignetting in a super subtle way

    Ah, vignetting... that good old friend that ruins everything. 

    At first, vignetting was just an artifact of poor camera and lens quality that we notice with the darkening of the corners of your camera sensor.

    Though, some photographers like to play around with it because darkening the corners of your image help in focusing your attention on the actual subject.

    The thing gets complicated when people take it to the letter and deliberately overdarken the corners of the image so that the subject is "clearly emphasized". 

    Well. It doesn't work this way, and by doing so you're just ruining your image.

    If you want to use vignetting, keep in mind that it is rather a layer of realism you can add to get a kind of retro / crappy camera look on your image rather than an actual composition tool. 

    If your composition is not good in the first place, overplaying the vignetting tool won't help. But if your image is well balanced, a really (really) soft vignetting can be interesting. 

    Also keep in mind that vignetting is an artifact that comes from the decrease of light, which means the amount of vignetting can vary from one corner to another on the same image depending on the amount of light you receive in this corner. Concretely, a corner of the image with the sky will receive more light, thus have less vignetting than the opposite corner of the image with, say, grass.

    Use DOF when deemed necessary

    Depth of Field effect is a pretty cool camera setting that enables you to focus your camera on a specific distance and blur everything that is beyond and before that distance with more or less intensity depending on the aperture of your camera. 

    If you use it directly in your 3D renderer, the good thing is that you'll have something physically accurate. Then the only thing you have to be careful of is just to actually use DOF correctly so that it actually brings something to the table.

    I'd always suggest being subtle and measured when using DOF, even if you're doing close-ups, because it can quickly become overwhelming and actually ruin the overall image. 

    If you intend on making DOF in postproduction, be extra careful, on top of not ruining your image with a poorly used effect, to mimic a physically accurate and plausible DOF (use Z-depth pass for instance).

    Soft DOF that focus the image on the foreground vs. unrealistic DOF effect that just makes you dizzy

    Homogenize your sets of images

    Most of the time, your clients are going to ask for multiple images for the same project. 

    Different scenarios can occur (only interiors, only exteriors, a bit of both, close-ups), but overall the idea is to keep a certain consistency in your set so that we understand that all these images were done by the same person and depict the same project. 

    Whether its contrast, light or hues, it is important to have a certain amount of homogeneity in your whole set so that once all the images are put side by side (generally what happens for a presentation where all the panels are printed) there's no mismatch. 

    The best way to achieve a good balance in the overall series is to create a new document with all your images placed next to one another. Then you use our typical check filters (black and white, saturation and hue). You then tweak each aspect for each image until you achieve an overall balance in the whole series. You can then copy your adjustment layers for each image back to their original file to apply them to the final high resolution file. You can also add a ready-made filter (Nik Collection for instance, or your own LUT) with a really low opacity on all your frames.

    From top to bottom : original set, original black and white, original hues, corrected hues, corrected black and white, corrected set.

    That'll be all for the pieces of advice I can give you to start the year fresh (with about 20 days of delay). Hope you'll stand by some ideas in the list for this year to come.

    And you, what are your new year resolution as an archviz artist?