The theory and flow in Image Manipulation?

[beefhitler]

How exactly is a typical image manipulation done start to finish IN THE BEGINNING if we are not shooting anything ourselves but dealing with stock images?

Let's say I have a concept in mind. I have a sketch in mind. And now I need to put these things together.

Example, I want to create a scene where a car is driving down a desert highway and it's being chased by a helicopter. There's an explosion in the far background. This desert highway is right next to a coastline.

Go out scouting for images on Google.

I have downloaded everything and then on a blank layer, lay these images on top of one another to make the rough outline of my scenery. Looks messy without the masks and all but I know this is gonna look great later.

And then I put the car on and it looks like the perspective is all off and if I do work on correcting the perspective, it looks like a badly drawn car.. kind of cartoonish. Seems like I have to throw this car away.

Keep looking for more images on Google for a car but no matter what, the perspectives of the cars I found are never going to be easy enough to work with to fit my actual composition. Probably spent about an hour or two just doing this. Cars are difficult for me.

Then I figured out, if I had started off with picking out the car first, and then look for the other stuff like the desert, sky, sea, coastline, hills, helicopter etc, it would have been easier. I now have something to begin with.

My entire set of background images were picked based on the extent to which I could manipulate the perspective of the car without making it look horribly off.

My next step would be to do the actual cutouts, detailed masking, matching color and light, light painting etc with all the images.

Lesson learned here is: Always pick the most difficult image first while picking stock images. It is usually the objects in the front of the scene. In this case it was the car for me. Then it was the highway. I find it easier to match perspective on natural elements like mountains, hills etc. In short, get the background images later.

This is MY theory and I am a complete noob with Image Manipulation. I think I am wrong as I don't have the experience. I want to learn it in the proper manner and not spend several hours trying to perfect something which I am approaching completely wrong.

What exactly is the process followed by you guys for these kind of stuff?

How would you have done things differently in this case?

I am not looking for speed art or tutorials as there are already plenty everywhere.

I have noticed that almost every thing related to Image Manipulation is already diving into the part of masking, coloring, detailing etc where the stock images are already there. I have not seen anyone discussing how and why are those stock images picked out. It is pointless to have 1000 images of a car if they don't have the perspective which would allow me to put it together with the rest of the 5 stock images for the background I have in my library.

Are there any advanced articles, books or videos etc which actually go in depth into discussing stuff like this?

I am looking for a bigger discussion on the part which deals with scouting for images and putting them together before actually getting down with the rest of the stuff like masking, coloring, detailing etc.

[Shoku]

Re: The theory and flow in Image Manipulation?

It helps if you have high marks (IQ) in spatial acuity. But to make things easier in the software, it also helps to shoot all the elements yourself, from the same camera angle. You may not have the time to do that, but that’s how many compositors work, either in studio and/or on location.
I have done many composites using client supplied images and most of the time they have shot each element correctly. When they don’t, it complicates things but is usually doable with a little more time involved.

[hendrix5757]

Have you done a lot of experimenting or have a significant amount of experience using Perspective Warp? Just concerning the point in your post about struggling with finding an image of a car in the correct perspective seemed like an opportunity for bringing the Perspective Warp feature out to play. I'm sure out of the many stock pics you found online of cars shot from various angles, there would have been a few that could've definitely been warped enough to be convincing. One thing I noticed from watching several time lapse and tutorials from expert photo manipulators was their tenacity to forcefully modify, alter, and distort any stock images using every available tool at their disposal. As if they refused to take no for an answer, regardless of what kind of lighting, form, color, or perspective information that may have been baked in. Furthermore, the most significant observation that had a remarkable impression on me was how much a formal artistic foundation came into play when modifying a specific element. Watching them take a 'piece' here and a 'piece' there and blend them together demonstrated how little the software's utilities determined the overall quality of the outcome. Using their knowledge of lighting, form, and dimension to dictate where to apply blur, how to transform/warp something, or where to place a shadow/highlight clearly showed me how much work I have left to do. Knowing everything about the program will obviously provide an advantage as to how efficiently you can attain results, but that will only get you so far. It's all about marrying those skills - artistic sensibility and technical facility . Get in a habit of limiting the resources available to you. Creativity tends to spark whenever an artist's back is to a wall, and is forced to work under limitations. Try not depend on searching for an element that fits the exact specs of your photo, but figure out how to take an existing image that's close to what you need and TAILORING it to what it needs to be. Make do with what you have. Identify what needs to be done or determine what's lacking, and then apply what you know in order to get it there! Problem solving is where all the fun is! Take the time to go crazy experimenting with tools and tricks. After all, if none of the elements were captured in the same controlled environment or under consistent conditions, you can forget about making everything harmonize perfectly. Plus, those preparations are to ensure that everything comes together as seamless as possible during post, but as you mentioned, this is a situation without that luxury. You'll learn so much more by conditioning yourself to react spontaneously to situations that demand artistic ingenuity than you will attempting to adhere to some formulaic procedure. I'm aware that these suggestions are more theoretical and unconventional, especially compared to many of the straight-forward methods you'll come across to guide you in creating professional looking composites. I was aiming to shed some light on the general skill set that needs to be acknowledged and nurtured in order to compose convincing photo manipulations. Much of this may seem like common sense or painfully obvious pitter-patter, but many professionals can attest to how often retouches fall short due to a visual lack of adhering to the basics. Now I know you said that you're not really interested in watching speed art and tutorials and such, but please check out Alex Koshelkov's videos on youtube (if you haven't already!) with the intention of observing how he resourceful his overall approach is. He doesn't use layer masks, doesn't name his layers, or use smart objects. He just goes for it! All his videos are terrific demonstrations of artistic and technical flexibility! Anyways, I know this is over a year old post, and I'm sure you've climbed leaps and bounds since this initial post, but I hope this helps a little.