Jump to content

Your opinion about 2D people integration in renders


khaledbou
 Share

Recommended Posts

Hello Archviz community,

 

I'm practicing post processing, inserting 2d people in renders,

I've used evermotion archinterior scenes for for this, I want to achieve a look as natural as possible.

 

for overcast weather & soft lighting it's not hard I think, the real challenge is in the sunny days :p

again casting shadows on ground is Ok but trying to determine where shadows are cast on other people and near objects takes skill (or maybe there's a tool for this I'm unaware of ?)

 

Also, shadows and direct light are not the only thing to take onto consideration, it's also about indirect lighting (GI), ambient occlusion, reflections, perspective, scale, having the correct blur, story, and of course good masking...

 

I know that 3D people can spare you from all of this , but the ones available today are either too expensive or not realistic. plus you need to know rigging in order to pose them in an interesting way.

 

I'm thinking about taking tutorials from photographers like Phlearn to learn correct compositing...

 

could you please critic this works, and share your workflow of populating your renders.

 

01.jpg

 

02.jpg

 

03.jpg

Edited by khaledbou
Link to comment
Share on other sites

I stay away from 2d people as much as I can because you really have to spend a lot of time getting them to look right. The examples you've posted illustrate why this is so difficult, the lighting of the people is different from the environment and this causes them to stand out like a sore thumb. 3D people have gotten much better over the last few years and while they are more expensive than 2d people I think when you add in all the time it takes to populate a scene with 2d people you'll find they really aren't that expensive after all.

Link to comment
Share on other sites

While yes the lighting can really mess you up, I think my biggest issue with the first two are the placement of people. You packed so many in there that I can no longer see the space. The third one looks the best, but it looks like a combination of better lighting on the 2D people as well as better placement.

Link to comment
Share on other sites

Thank you much guys for your great feedback.

 

here is two renders for comparison,

Above : free 3D people used from AXYZ and renderpeople.

Below : 2d people used from different sources (+ my average compositing skills :p)

 

3_D_people.jpg

 

2_D_people.jpg

 

 

My opinion: 2D people are the best.

because even if you're whiling to spend +1000$ to build your 3D people library (and a new hardrive), there's simply not much variety! you may need people in different situations, doing different activities, their clothing must be in different seasons.

 

and about time consumption, It's really the same thing; with 2D people you search for pictures en google or use from your cutout library, mask them, and adjust them to your render.

with 3D people, you may need to do the rigging, if they're already rigged you need to pose them, you may also need to change their materials to have variations, and only then comes the many test renders until it looks fine...

 

 

Plus: if you want to see the true potential of 2d people, check these two :

https://www.beautyandthebit.com/work/

https://mir.no/work

Link to comment
Share on other sites

I do not think 3d people are anywhere near in terms of quality that you could get away with. For distance shots we will sometimes use them but when i say distance i mean 100 meters away or further. Everything else gets post productioned in using photographing images. So, specifically talking about your images:

 

1. Inserting people into images ir real art and takes years to master.

2. You are inserting people that have been photographed outside in sunny conditions into an image that has pretty ambient/even lighting with no strong light source. What do you think is going to happen? So the choice of your people is wrong. That is 50% of your job that you are getting wrong straight away because of the choices that you are making.

3. Color correction/tone is extremely important. For example, image number one has a soft warm tone and your people need to have some of that in them as when the light bounces around it would effect them as well.

4. Shadows- that is what makes people "sit" well within the scenes. If you don't get that right your people will stick out like a sore thumb and the best way to get it right is to google spaces and have a look at what happens in reality. Never paint shadows the way you think they happen in reality!!! Observe in photographs what happens and then try replicating.

5. Composition-this is what makes or breaks the image. Your people can be well fitted into the scenes. Correctly lit, with a perfect tone and perfect shadows but if you cannot place them well it just won't work. For me that is one of the most important parts. This is what separates ok CGI artists into great ones. To balance an image and understand what it takes to get there is real art. There are rules to a good composition and you can find all the relevant information online. However you only become great when you understand how to break the rules and push the boundaries to greatness. I know it is vague but you can view the work of the usual suspects like pixelflakes, MIR, vyonix etc. The all have their own style but the all get how to get an image over the line to make it awesome.

 

Hope this helps...

 

All the best

Link to comment
Share on other sites

  • 2 weeks later...

I think you have 2 issues.

 

1. The lighting of the 2D people is wrong for the lighting in your shots. Most of your interiors have very flat / diffuse lighting (little or no directional light), but the 2D people have been shot in the sun light. Try to find 2D people who have been shot either in interior spaces, or on overcast days.

Then you need to look into compositing the people into the shot better. Adjust the brightness and contrast, matching blacks, etc.

 

2. The placement of the people is just not "photographic". On your first image you have people walking directly towards the camera, and the focus goes straight to them, especially the dad and child. And what is the lady in the red skirt doing? Also think about groups, and where people might actually be going.

Think about how a photographer would shoot the space. They would wait until the ideal opportunity to shoot, not just shoot any old picture with random people blocking the view.

 

Dean

Link to comment
Share on other sites

Here are my rules for adding people to scenes:

 

1. Think of where your render is. If your render is in warm tropical weather ... DO NOT put people in sweaters and jackets. It is weird.

 

2. Use 'height boxes' in your scene. Create a box that is 1.5' x 1.5' x 5'10". Spread them through our your scene where people would be standing. Screen shot your viewport and paste it into your compositing software. Scale it up to match your scene ... bam ... instant perspective and size reference.

 

3. DO NOT use the following: people facing camera in foreground, people who are obviously posing for the photo, too many people in foreground (distracting).

 

4. Color correct your people (this includes painting shadows and highlights if necessary).

 

5. Use correct shadow colors and direction. How often have you seen a render where the shadows were NOT the same color as the 3D shadows that are near your person you are placing? Looks bad ay? Yeah. It does.

 

6. Have your people engage the architecture. Open a door and have your entourage walk through it. Have them pulling out a chair or observing a water feature ... something. This helps melt the 3D with the 2D. This is the one I have the most difficulty doing.

 

7. This is something I do if I have more time than usual for adding people. Create a story with your people. Why are they there? Who are they talking to? Are they mad? Happy? Are they waiting for someone? Eh... things like that.

Edited by digitalputty
Formatting
Link to comment
Share on other sites

I agree with much of what has already been said. The two big issues are that the lighting of the cutouts don't match the space and the busyness of the images. The lighting issues are common, but can easily be remedied by first finding companies which offer multiple lighting options, such as direct and flat lighting. You will still need to do some touch-ups, but it makes a drastic difference. For the amount of people you place, in addition to what has been said, remember the focus should never be the people. For each of your images, there is a person who captures my attention first and the space is secondary. Part of that is because they're brighter, but also because they're centered. Take your second image for an example, I'm immediately drawn to the white shirt. If he is removed, my focus may be on the space more. Larissa provided great insights on shadows, which is also crucial. Lastly, I think one issue that hasn't yet been addressed is the camera angle of the space. Using your second image again for an example, there is far too much foreground that the image seems to ask for a lot of people. Is the rendering itself supposed to focus on the stores or the open space? If it's the former, you should bring the storefronts closer. If it's the latter, you could adjust the image to give the impression you're walking out into the open space from a store. This can provide an interesting perspective, but also limit the number of people you need to include.

Edited by vizutilcompany
Link to comment
Share on other sites

One other tip is to lay stuff like a photo filter over the whole image or even some general PS noise (I know we spend so much removing noise but don't be afraid of a fine layer of uniform noise) that will serve to blend your composited people in with the whole 3d scene. So when you've satisfied yourself you've finished matching highlights and shadows of your people with your scene - do a Ctl Alt Shift E to snapshot the scene then smash some general filters or noise over the whole thing to create some uniformity of colour cast and overall grainyness. The eye really picks out the subtle differences in resolution, noise and colour cast otherwise. Overlaying something that affects your whole finished and populated scene is a quick band-aid to remedy that.

 

That's if you're renders come out of the frame buffer requiring minimal post work... if you're doing a lot of post work, just composite in your people early in the post processing so that a lot of the work you do on the general colours and mood of the scene are affecting your 2d people too.

 

But the reality is.... there ain't no easy way to get a really good result. When someone cracks a quick fix for this problem they'll make a fortune!

Link to comment
Share on other sites

Hi,

 

a couple of years I ago I wrote a quite lengthy article regarding seamless integration of cutout people in renderings. I've updated it a bit lately, taking into consideration how 3D people are now looking better (photogrammetry and all), still I rely 100% on 2D cutout integration for my daily renderings.

 

You can check the article here : http://www.horoma.net/tutorials/how-to-populate-your-render-seamlessly

 

The tedious part about inserting cutout is generating a full set of layers to get hue, saturation, values right. Then get directional shadows on all your cutouts. This is why I also wrote a script that creates the whole structure for each cutout, neatly, in your photoshop folder.

 

Hope that helps!

 

Cheers

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...