Using AI to Make Composites : Firefly, MidJourney & Generative Fill in Photoshop Beta

welcome back everyone in this week's video I'm going to share with you how I've been using AI over the past few weeks to generate these composite images of NFL football players so I'm going to go over how I used vato AI to retouch the skin I'm going to go over how I used Firefly and mid-journey to generate the backgrounds how I use generative fill to fix those backgrounds and then I'm going to talk about how I use Photoshop in order to blend it all together I'm also going to talk about how I lit all of these photos as well and sort of what it's like being on a set like this so when I was a teenager I collected trading cards and that eventually led to me going to get autographs from professional athletes and then one day I thought maybe I'll borrow my mom's point and shoot camera and take photos of them and then I'll print them out and have them autograph them the next time I go to one of their games and that's actually one of the things that led to me becoming a photographer so when I was asked by a trading card company about eight or so years ago to photograph um NFL draft picks and then NBA draft picks for their rookie trading cards I jumped at the chance because even though I don't really pay attention to sports uh now I know that the child inner child in me would have loved to have done it and it's a lot of fun to do one of these shoots because what I end up doing is I end up photographing about 40 different athletes in the same day from that Sport and I'm working with an art director in order to get all of the poses that they need for the trading cards and this year what I did was I decided to mix things up a little bit and do five different setups on the same set in 10 minutes and I did that by using the Ellen Chrome mobile app and then controlling up to eight different lights so I could turn them on or off or up or down so for the first setup I did this clamshell lighting setup and I had a strip soft box in the front vertically and then at the beginning of the shoot I had a small a light with a seven inch reflector down below and I was using that as my fill over the course of the shoot I started to feel like it was a little too much fill and even though I had it at minimum power and I was using an ND filter I just decided not to use it anymore so I took that out and just went with that one light vertically over over the top so it was a clam shell in the beginning then it was only one side of the clam after that then I let the players from the back with two strips off boxes one and over each shoulder and a hair light from the top center of the backdrop then I repositioned myself over to my left and had the athlete turn over towards me and now I sort of started to shoot instead of on the backdrop I started to shoot onto a black curtain and for this shot I used the vertical strip box that was on in the front for their face and then I used one vertical strip box over their shoulder in order to create an edge light and this was to follow a setup that the art director had that he wanted to do uh for all of the trading cards and so then for our third setup I had the player stand in profile and I lit them from the back with the strip softbox over each shoulder and then the hair light and I had them face both directions and look at different areas along the wall in order to get the right eye line that I was looking for then it was time to shoot the action so I decided that we were going to do some really edgy shots and we did that by using a 70 centimeter or 28 and a half inch OCTA box over here on camera right sort of equal in depth from me to the background in line with where the model was going to be running and the art director had them go through the plays depending on which position they played he would have them do different things and that's how that worked out I also lit him or lit the players from the back with a strip softbox on each side and the hair light from above then to get a less Moody action shot we turned off that 70 centimeter strip softbox and we used a 120 centimeter light motive softbox over on camera right for our main light and we used a umbrella on camera left as our fill now the reason why I use this round light source is because I did not want a octagonal catch light in their helmets I wanted it to look like it was maybe a stadium light or something like that so that's why I'm using that light source in particular now you may have noticed on some of the BTS photos that the light mode of 120 over there on camera right and the umbrella on camera left were pointed back towards the camera and not at the set and the reason why is because in those shots they weren't in use and I didn't want them to act like two giant white reflectors so I turned them around now some of the time you're going to see that they're firing and they did that but they were turned turned down and it wasn't affecting the shot the way that I use the mobile app was to Simply program in each of my setups including the power settings for each of the lights when it came time for us to move between looks all I would do was grab my iPad and load the configuration file for that setup and it really worked out great now almost in real time I delivered raw files to the client so they did all of the selection for their own purposes for however they were going to use them I however decided that I would pull through the images and come up with my favorites so that I could use those to make images for my portfolio and that's what I'm going to be showing you today these aren't images that I created for my client they're images that I created for me because I wanted to learn how to use all of the AI tools to make this happen and also if I posted 50 photos on a gray backdrop that wouldn't be very interesting now would it so that's that's how we got here to this point so once I called through and I found my favorites I just processed them and Capture One in a very straightforward manner with the base characteristic of high contrast and I did this because I wanted a very Punchy image which is sort of what I think fits in well with this athlete sort of genre maybe I pulled down the highlights sometimes maybe I pulled up the Shadows sometimes but just know that I wasn't doing any heavy color grading I was just taking the images as they are and processing Tiff files then I opened up those images in ivato Ai and if you're interested I'll put a link up here to a video that I did a few weeks ago where I talked about that software specifically so the way that I use that software this time was I just wanted to remove all the blemishes on the face and the body I wanted to brighten the dark circles around the eyes and remove some of those wrinkles a little bit but I didn't want to go too far I also did a little bit of Dodge and burn on the photos and I did all of this by by setting up a preset in the software and then I would apply it to the images and it would take about one to three seconds for it to process so it's a very very quick sort of thing this software is great for batch processing it was also great for something like this where I needed to retouch them but I didn't need to retouch them and make them look too pretty I just needed to generally polish the images and that's what it's good for now it probably I know it did from time to time it missed some blemishes on the face and I would have to go in and clean those up but it did 99 of the work if you were retouching a Beauty image it might get you to 80 percent of the way there and then you're going to have to come in and clean it up by hand and that's why I use the software for something like this but I still send images out to a retoucher to get those done for my normal work for these pictures I did it all myself using the software and this process I'm going to mention because I felt it was something that I wanted to learn and part of the artistic process then I started generating the backgrounds or the background plates that I was going to use in Adobe Firefly and mid-journey and Adobe Firefly is completely free if you already have a subscription with them for Photoshop and Lightroom I believe it might be free otherwise if you want to check it out because right now they are beta testing I found that Firefly was really good for creating heavily bocaed images of cityscapes like this one and it was also good for generating Landscapes and so let me share with you the prompt that I use in Firefly to get this image that I use for the background here so this one was Malibu Beach rocks with palm trees at Sunset warm tone backlighting bokeh Sunset I guess I did that one twice Orange Sky 35 millimeter clouds so as you can see it's a lot of descriptive keywords that are in there but the key part is sunset that orange Sky clouds backlit that way you can get the light coming in towards the camera and that'll give you a motivation for a kicker or at least that's what I was looking for so I used the sky brightness to make it make sense for there to be kickers or Edge lights in the shot so that's how that one worked now mid journey I found to be a lot better at creating scenes like the stadium scenes that I used or the tunnel that I used in today's tutorial so the keyword that I used for this one was an american professional football player stands in an old blue painted concrete pedestrian tunnel with lights on the left side only full perspective waist up at night backlit lights cinematic lighting medium shot 35 millimeter F2 now the reason why I had all of those commands in there origin was because I was trying to create an image for this Detroit Lions player and I wanted the light to be coming from one side to give motivation for the kicker I just ended up setting him off to one side and then that made sense but when I was creating those variations that's when I got this image that I used for today's tutorial the reason why I was using an American football player as a prompt was because I wanted something in the shot to be focused that way the background being out of focus and falling off as it went deeper into the shot that would make a lot of sense then I would open up the images in Photoshop beta and select the player and then use generative fill without a prompt I would just hit generative fill twice and it would remove the player from the image then I would merge that layer down then the next thing I did is I would open up my retouched image into Photoshop beta I would then crop it to 16 by nine thus expanding the image to a horizontal even though it was a vertical and so then what I would do is I would drag and drop the same file into the program again and then that would be a smart object and I would link that to the retouched file the reason why I did that was because if I was going along in the process and I decided that I needed to reprocess the research file again I could just re-link this image to the new retouched file or if I had the same file name it would just automatically do it then what I would do is I would go back to my background image now unfortunately these AI generated images are not very big they're sort of close to a thousand pixels in length now they're a little bit bigger than that but they're not very big at all and so sometimes you've got to upres them so what I would do is I would open them in Photoshop and I would go to filter neural filters and then I would go to super zoom and I would press the enlargement button to about 8X and then I would tick those first two boxes and then I would set the sharpening at 15. now this worked out most of the time when I was using a background plate that had a lot of detail in it but when I was using an image like this one that we're doing the tutorial on today or the breakdown on today nothing in the shot is really sharp and so when I use this neural filter process to upsize it it actually made a softer image than if I just use transform and increased its size that way so I'm going to paste my background image here and then expand it to get the exact size of image that I want to match the foreground image I'm going to select the top layer and then I'm going to hit remove background I'm going to click on the mask and then I'm going to go up to select and mask so I can start to refine things if you look over here on the right I've Got The View mode set to Overlay and that's so that I see a red background on What's Missing and then everything that's in my selection is showing up here in the foreground now normally I would use the refine Edge brush to fix the hair selection but this hair selection actually looks pretty good now if I didn't look good I would use the tool right now to get that squared away I'm going to set my output here to New layer and hit enter the next thing I'm going to do is I'm going to zoom in here I'm going to go with a 100 hard brush and I'm going to click on the mask and I'm going to go over the areas that I need to add I'm going to add them with a white brush and I'm going to go over the areas that I need to subtract and I'm going to subtract them with the black brush now I'm going to zoom in here to his hair because I've learned a new way to clean up the edges I'm going to switch back to White I'm going to change my blending mode to Overlay now what I'm going to do is I'm going to paint along the edges of his hair and it's going to darken those parts that were a little translucent and make them better make them look better in the selection if I go a little too far and it starts to select some of the background I'll just switch over to Black and then clean up those edges but this makes a much better mask overall the next thing I'm going to do is I'm going to use the neural filters to harmonize the color and contrast in the image so I'm going to go up to filter neural filters and then I'm going to click on what else but harmonize then over here on the right since I've got my top layer selected I'm going to tell it to sample from the background layer so I'm just going to drop down here to that then what it's going to do it's going to analyze that background layer and decide how it should change the foreground layer so that it will have the same contrast and color tone by default I believe it's going to come in at power setting 75. you want to season this to taste maybe it's going to be a little too strong maybe it'll be a little too weak I haven't seen that actually it's usually always too strong and for this image I'm going to scale it back to 50 and hit enter now the next thing I'm going to do is I'm going to make a new layer by holding down shift command and in I'm going to call it blue edge lights then I'm going to hold down the option key and I'm going to click on the mask of the player and drag it up to this new layer this will copy The Mask I'm going to grab the Color Picker here and grab some cyan from the background and then in with my paintbrush I'm going to paint that cyan color over all of the highlights in the shot then I'm going to double click on the layer to bring up the layer style adjustment window and I'm going to change the blending modes from soft light to Overlay to color and see what works best for me on this one I'm going to go with overlay then the next I'm going to do is I'm going to change the blend if here at the bottom I'm going to grab the black slider on the bottom and I'm going to pull it to the right this is going to restrain my blue paint to only the mid tones and highlights based on what's in the player layer then to feather it I'm going to hold down the option key and click on that slider again this will split it and then I'll pull that part on the right further to the right and hit enter this will make it a lot more gradual as that blue color transitions there from the highlights to the mid-tones on the player and guess what I forgot the player's name because even though the teenage me would have known this the adult me doesn't and I had to look it up at his name is Bryce young now many of you are going to be making fun of me for not knowing that but this is me and the reality so sorry uh the next thing I'm going to do is I'm going to add a new layer and I'm going to call it grain then I'm going to go to My Color Picker and I'm going to set the L value to 50. then I'm going to grab a hard brush and I'm going to paint over this entire layer in order just to create a gray solid then I'm going to change the blend mode to Overlay next I'm going to go to filter noise and then add noise the reason for this is I'm going to unify the image a little bit by creating a grain pattern over the background and the foreground which hopefully to the eye will make that read as though this is a legitimate image the next thing I'm going to do is I'm going to make a new layer and I'm going to call it cyan then I'm going to go back to the Color Picker and I'm going to grab a cyan tone here from the background then I'm going to paint that cyan color over this entire layer and then I'm going to change the blend mode to color I'm going to start off by setting the opacity to zero and then I'm going to pull it up ever so slightly so that I can see that cyan color sort of go over or wash over the entire image and here around five percent looks good to me and it just unifies the color a little bit more the next thing I'm going to do is I'm going to add an exposure layer adjustment then I'm going to pull that exposure value down about one third of a stop then I'm going to go over and grab my black paint brush I'm going to set the hardness to zero and I'm going to paint over Bryce's face in order to take him out of this adjustment layer so effectively what I'm doing now is I'm darkening down everything in the shot except for his face and that's going to view and that's going to draw the viewer's eye back up to it now there's one thing I want to do in order to polish this image just a little bit more I'm going to make a new layer and I'm going to call it catch lights then I'm going to take my white brush and I'm going to paint over with a very small hard brush the catch lights in his eyes then I'm going to change the blend mode to Overlay and I'm going to use my Eraser tool to sort of pull down the brightness here on his right eye just a little bit and this will greatly brighten those catch lights and even them out and make it so that the viewer will see them when we're zoomed out now you might be wondering one thing why didn't I use generative fill to make the backgrounds instead of mid-journey and Firefly well I found right away that it would take about a minute or longer sometimes for generative fill to come up with that background image and it really wouldn't be what I was looking for and even though it gave me three options they weren't that great I could do the same thing in Firefly I could generate four images in about 15 seconds so it was a lot easier and better for me to do it that way of course doing it with generative fill wouldn't give me options access as well to all of the mid-journey options as well so that's why I found doing it this way was better it is sort of a drag that these images are so small that I'm using to make the backgrounds I'm sure they're going to improve over time maybe a year from now we'll end up with 4K images maybe a year from now that generates it fill thing will work for the backgrounds and it will do it quickly who knows what's exciting to me is that we've got so many more options to try and so many more things that we can do to execute our Like A Creative Vision and that's what I love about it just so you know I wasn't paid by any of these companies today to make these this video I just wanted to share with you what I've been learning over the past few weeks in hopes that it's something that you might want to try anyway guys thank you so much for your time if you could please like And subscribe and sign up for the bell and all of that good stuff I would really appreciate it and if you'd like to watch a video that YouTube thinks you'll love click on this one up here and if you'd like to watch some of my lighting tutorials click on this one thank you so much have a great day and I'll talk to you soon

Channel: John Gress

Views: 92,861

Keywords: photoshoot, headshot, profoto, elinchrom, lighting, studio, portrait, photographer, photography, tutorial, godox, how to, light, octabox, alien bees, retouch, retouching, photoshop, skin, frequency separation, Lightroom, LR, lrc, adobe, post, production, colour, ai, photo, editor, processing, tools, Evato, Mid, Journey, jorney, fire, fly, fill, neural filters, super, zoom, superzoom, adjustment, layers, blending, select, mask, extraction, refine, hair, Bryce, #1, number, one, nfl, draft, pick, mobile, app, football, player, genarative, Young, Sports

Id: aawE1ZFxl78

Length: 20min 52sec (1252 seconds)

Published: Mon Jun 26 2023