Behind the scenes of our RDRFR preview

Now that we've officially announced and published the first preview of RDRFR, we have the opportunity to share a unique perspective of things with our Patrons.  One of the perks that we've promised is a more 'behind-the-scenes' look at our work, and I thought perhaps it'd be interesting to give some insight as to how exactly we come up with these previews, and what sort of work it actually entails.

A lot of the time when we're showcasing features or concepts through screenshots, it's very difficult to convey what exactly we want to show through one picture alone.  In some cases, like our PIT maneuvers for NPCs in GTA V, there was simply no option but to record a short video and show it in action.  For the most part, though, we want to share screenshots as they're easier to display and we don't have to worry about embedding them in posts, formatting issues, and where to host them.  Using screenshots also lets us save videos for the stuff that is truly special, too.

In the case of this first RDRFR preview, we wanted to show some of the initial work that we've done on the mod, but one of the challenges that we often face when taking pictures is getting the trade-off between the picture looking good (e.g. camera angles, depth of field, etc.) and actually showing the features of the mod - especially when they involve HUD/interface elements.  Now, in this first preview of RDRFR, one of the things that we wanted to prominently highlight was our ability to create contextual prompts in exactly the same manner that the game itself does.  To arrest someone, for example, you hold down the key and the progress indicator around the on-screen prompt fills up.  The problem we have, though, is how do we convey this when the Photo Mode (used for taking screenshots with free camera angles, etc.) completely hides the game's HUD.

In the picture here, you can see the saloon scene from our preview from the actual gameplay perspective.  Basically, there's been a bit of a brawl - to put it lightly - in the saloon, and the player is questioning one of the few remaining men standing, using the contextual actions we've added to RDRFR.  It's not the best picture, though, as it is taken from the gameplay perspective and as such there's a few things that aren't that clear.  

First of all, it's not really clear that it is the player who is speaking the dialogue.  You can see the dialogue is subtitled, but you don't really know who is saying it and this is furthered by the lack of highlighting around the prompt option for 'Question'.  In the game, when you press one of the buttons (F, R, or E), the button is highlighted for a second to show that you've pressed it.  Here, however, I have taken the screenshot a second or two after pressing the button and hence it's no longer highlighted.  This isn't ideal, as it would obviously be clearer that it is indeed the player saying the speech line - and therefore an actual RDRFR feature - if the button press was highlighted.

The second thing is the actual camera angle itself.  It's maybe not that clear that the player is a deputy sheriff, because you can't see the badge on his chest.  We can get around this by using the Photo Mode and playing around with the angles, but that means we lose the prompts and HUD altogether, which basically defeats the whole purpose of the screenshot.

To get around this, we use a little trickery.  We take three more screenshots.  The first is the cinematic camera angle that we'll use for the actual preview picture.  It's taken from the front, with the sheriff star clearly visible.

You'll notice, though, that we no longer have the dialogue subtitles or the contextual prompts, and this is where the next two screenshots - and the trickery - comes in.

First, we take a screenshot of the subtitle on its own, against a clear, solid background.  The easiest way at the moment to do this is to simply point the camera at the sky, change some of the weather settings to minimize the cloud coverage, and then display the subtitle.  After this, we then do the same with the prompts. 

When both pictures are put together, it ends up looking like what you see here.  I've been careful to make sure I hold the R key as I take the screenshot, so that the prompt is highlighted, and everything is against a relatively clear and solid background.

Obviously, though, this is of absolutely no use to us in its current form as it's merely a very weird looking picture of a floating person with some strange subtitle text clearly pasted over it, so time for the trickery.

There's likely other ways, but my favourite method for doing this is actually using video editing software - Adobe Premiere - which has fantastic keying functionality.  This is basically the same as 'green screen' from the movies, except in this case we've got a blue background and not a green one.  We simply select the background color with the color picker, and then let Premiere do the rest as it aggressively filters out subtly different shades of blue from the image.

As you can see, the end result (after a bit of cropping, too) is the goodies and nothing else.  We now have the subtitle and contextual prompts, with none of the background.  We can then easily overlay this on the second screenshot that we've taken (the one from the Photo Mode) to achieve the final result, which is shown at the end of this post and is the same screenshot as posted on our preview.

Now, while this does employ some trickery - it isn't deceptive.  It's not like we're adding anything that isn't already there - we're just taking the best of both worlds and compositing them together into one picture.  We want to take a cool shot, with the ability to change the angle and such, but we also need to display the HUD and subtitles.

The end result is that we get both, and I think it really helps to convey what's actually happening in the picture and what it is that we're showing off.  When you look at this picture without context, it's hopefully much clearer that the player is questioning the stranger, and this is exactly what we're wanting to achieve.

Then again, if you're to tell me this is way too much effort for a game mod, I wouldn't disagree!

- Sam.