I keep saying I am not going to do any more AI images, but then I keep trying out new things that Adobe is adding to their Firefly program to see exactly what it will do now. (Click if you want to try out Firefly – it is free for Photoshop users as the Generative Fill command is based on it.) It is really fascinating! There is a new Reference section that lets you use any Reference image to help generate AI images. Colin Smith of PhotoshopCafe fame (and AI expert) made a short video on exactly how to do this called New Adobe AI Feature Changes Everything Again. His results seemed to be a little better than mine – I found it took many iterations to get something that I liked. My feeling is that Firefly still does not create small objects that well – lots of clean up was required. I will just give you a quick recap so you can see what I mean.
The above shows my Original image that was uploaded into the Reference section. (I am still uncomfortable using other people’s work in AI.) The prompt used to generate the Firefly images was “Venice canal with gondolier and bridge.” The Base Firefly image only used the Structure Strength at 100% and Content type set to Art. The Tower image used the Reference Structure set to 50% Strength; Content type set to Art; Style section with uploaded using my Original image again at 50% Strength; Effect set to Oil Paint; and Color and Tone set to Golden. The Moon image used the same settings as the Tower image except the Style Strength was set to 0%.
According to Colin, once a Structure image is uploaded to Firefly, a Strength slider shows up. By increasing the slider to the right, the generated result will look closer to the original uploaded image. The Strength is just picking the basic outline – the object, size and positioning. This can be seen in the results obtained in all the generated images. When Strength is set to 0, the image is not as close. Use the Styles section to change that part of the image.
Above are examples of other generated images that were not selected and used different strengths and effects. They look pretty good at a very small size, but if you look closely, there are lots of problems. One of the biggest issues with the Base Firefly Image was with most of the building windows – they were all turned different directions, were different sizes, and totally out of perspective. Had to use the Remove Tool to get rid of all but the ones I liked, and then Clone them into other areas to get a nice look. Also had to clone the posts on the bridge so they looked balanced. The gondolier’s arm was not connected in any logical way to the oar. Totally weird and this happened in most of the other generated images. Also a lot of the hats looked really strange upon close inspection.
I guess the bottom line is that this is by no means perfect yet as it appears to be in other AI programs. Be ready to do some massive clean up if using it. There were several layers of painting and cloning in my image above. And my original gondola was added with a sketch line placed around it. It was fun to see what Adobe Firefly is doing to improve their program, but it still has a long way to go. Have a good week!…..Digital Lady Syd
Wow! Where to start with this one! I learned so much about the 1920’s fashion era just by researching the information needed to create this fun Rebelle 7/Photoshop image. I have to thank Corel Master Painter Karen Bonaker of Digital Arts Academy (she has the best free community and classes for anyone interested in digital painting – check out its header) for suggesting to her members to try a Gustav Klimt project, using some special brushes she supplied. (Will talk about below.) I imagine almost everyone knows what Klimt images look like, but here is a link just to refresh your mind. Also, this image could have been done completely in PS if you do not want to purchase Rebelle 7.
My subject’s appearance is a bit like a1920’s era Flapper with her short-cropped hair and bright makeup, but the dress, shown in the screenshot picture on left below, had long sleeves (not the Flapper’s short-sleeved look) and the fabric was very soft, flowing and elegant. The more fashionable women wore the “Great Gadsby” look based on the classic book from that time period. She is wearing a Cloche Hat worn from before the 1920’s thru 1933, which has an incredible history of its own (several are available on Amazon). The screenshot (apparently of silent movie actress Pola Negri, who had a very interesting life) was from a blurry AI enhanced and colorized vintage 1927 short video of the streets of Paris – it was up sized using Topaz Gigapixel. The sleeves on her dress were so pretty and seemed perfect for a vintage painting. In PS Pola was selected from the screenshot and placed on her own layer; flipped horizontally using the Transform Tool; Puppet Warped to get the arms right; and last, roughly sketched with Kyle’s Animator Pencil (2016) that can be loaded from PS. See sketch below on right. You can see she is not exactly like the actress shown in the video, but I wanted to keep her face and basic dress features similar.
REBELLE 7
The PS document was taken into Escape Motions Rebelle 7, the fairly new painting program that has improved so much it makes Corel Painter look pretty old school. If you watch for sales, it can be bought at a good price, but be sure to get the slightly more expensive Pro version which has all the really cool features. This program seems pretty easy to learn and has some really nice painting brushes to use. This is where Karen’s Klimt and Art Noveu brushes were useful for creating this image. First a background using the Klimt Oil brush Burnish Pattern was painted. Her Nou 3 in the Art Noveu set was used to create the circles in the upper left corner. The red scatter points in the upper area were from the Klimt Pastel Pattern brush. The Klimt Variable Texture brush at the top left was actually duplicated in PS and flipped and placed on the upper right. Since I am still figuring out the brush system, the image was brought back into PS where all my favorite painting brushes can be used on the image. But I still wanted to use some of Karen’s custom Rebelle 7 brushes in PS.
CREATING A BRUSH FROM A PNG OBJECT IN PHOTOSHOP
Since these Rebelle brush sets are basically stamped brushes, they could be converted to PS brushes by downloading her free Klimt and Art Noveu.png brushes from Rebelle’s website. Drag the brush png over to a New Document with a white background (and get rid of the Smart Object by right clicking and choosing Rasterize). Since black creates the mark and white is transparent in a brush, click CTRL+I to invert the image. If the white area is still kind of gray, just add a Posterize Adjustment Layer and move to a level where the white disappears. (Can also use Brightness/Contrast if needed.) Now use the Rectangular Marquee Tool on the png layer to select just the paint dab at the top of the png. Go to Edit -> Define Brush Preset and name. There is your brush! You can use this process on any PNGs with cool objects – just remember they will probably only make stamp-type brushes, not for digital painting. There are no settings associated with the stamp brush just made. For Klimt look, this can be very handy to use since he uses a lot of geometric patterns in his images. The brushes converted to PS from Karen’s Rebelle brushes were mainly from her Klimt set which included Variable Texture3 for the dots behind Pola and the Rough Oily brush for the little individual squares placed.
A few other brushes were also used in PS on this image. PST Twirls brushes were used for the lower right side circles and the twirls by her head (from 2006 (Wow – but Brusheezy and DeviantArt have so many – search for Twirls). The Flourish chandelier effect was created using Vintage and Grunge Element 2 by Katie Lynn (cannot find her 2017 set anywhere but there are lots of flourish brushes available also), and then I created a 12-px Hard Round White Dot brush set to 189% Spacing to add the little lights to it.
MY LADY LAYERS
The Hard Round White Dot brush was also used for the detail on her belt with a small Stroke layer style. And then just a PNG object was used for her rose on the hat from a for sale set called Anemone WC Texture Clipart by YouArtMatter. Of course all of these had layer styles or adjustments layers added to blend them in properly. This is why I have issues with the Rebelle program – it does not have all these features yet. One of the coolest brushes I have downloaded is on Deviant Art and called Pixelstains Lip Texture which adds the neat highlight to the lips. From the video you can see the dress was a plain cream colored dress – not exactly Klimt looking. Her dress was selected and put on a layer, then a Pattern Fill Adjustment Layer was clipped to the dress layer – used DCandies Luxury 3D Patterns at a Scale of 15 and set to Lighter Color blend mode at 40% layer opacity. Next a Hue/Saturation Adjustment Layer clipped on top of it (-26/+4/+19). Finally another Pattern Adjustment Layer was clipped using Resource Boy’s WC 671 pattern (a floral pattern) with a Scale of 12% and the layer set to Hard Light blend mode at 44% opacity. On top of this, another layer was clipped for creating the necklace – used Jessica Johnson’s (this PS expert has some great techniques for creatives) Floral Lace Romance Brush Var with a layer style. This can get complicated, but it is fun to try different techniques and figure out what looks good! To finish up the dress, some shadows were created to make it look like it was soft and flowing. On a New Layer Sam Peterson’s tip for adding shadows (the Layer was set to Multiply blend mode at 34% opacity) was painted on her dress and to parts of the face in shadow under her hat. (See my A Few Photoshop and Lightroom Tips and Tricks, Step 2 for info on how to do this – I really like his technique.)
Above is My Purple Lady image – it has a different feel to it but I like the color combinations. The only difference between the two images is a custom Curves Adjustment on top was set to Color blend mode to create the yellow tones (in the RGB field drop-down, individual color curves were adjusted). The above blog is basically what was done on both images, although a Color Lookup Table Adjustment Layer and a Curves Adjustment Layer were added on the top. It took a lot of time and iterations to get the look needed. If you like to try out different brushes and experiment with layer styles, blend modes, and adjustment layers, this is the project for you! Also below are various digital painting blogs I created using different painters’ styles that you might like. Hope you enjoyed a little of creative journey with this image!…..Digital Lady Syd
I have been experimenting with all my AI Selfies, mainly trying to see what I like or if I like the results after painting with different types of Photoshop brushes on layers above the AI creations. This has been a hard decision for me on how to use this new technology that is getting more and more popular in certain mediums. The Adobe Firefly and the Firefly Photoshop results are great for me to learn how to use new brushes and techniques, but I do not see any real practical uses for my serious photography – except for the occasional little fixes in an image. And it does appear my AI generated Selfies do have a certain look that sort of gives away the fact an AI process was used. There are so many other expensive AI software programs creating more incredibly realistic images and it is totally hard to tell it is AI. With that said, I believe you must keep up on the new technology if you want to understand what purpose it will have in your own creative expression. Hence my Selfies and what I am learning.
At this point, I am only creating with Adobe AI and using their results to learn how to paint trying different art mediums, hopefully leading to my own original works down the road. Therefore I am displaying a couple more Selfies (since it is my photo, I do not have to worry about who work it is). But first I want to share the info in the Note below that caused me a lot of confusion when generating AI images using less than 100% of the image (which you have to do to get any good results) when I began using all the different processes to do this.
NOTE: This is important to know when using different amounts of the Brightness setting!If you get reverse results (in other words, set to 80 and the image is wild looking instead of somewhat close to normal looking), go back and double click on the Quick Mask to open the Quick Mask Options Panel and change the settings to the other choice. For example, when using Dr. Brown’s action (I do like the fact that his action creates a new document for his AI results – check out my Dr. Russell Brown’s Painting AI Action Set to download) or Unmesh Dinda’s technique (more on this below), Quick Mask needs to be set to “Mask Areas.” You will know if you have it right if at a Brightness of 80, the foreground swatch looks light gray. When using Dave Kelly’s action, it needs to be set to “Selected Areas” (he tells you that in his YouTube text where the link for his action is located. I find this all very confusing so just beware of this.
The top Selfie was created using Dr. Russell Brown’s Painting AI Action set to Pencil Sketch and using the 60% of original action variation group. There were all kinds of variations created, some really strange, but in the middle of the batch, a halfway descent version showed up. Not sure it was really a Pencil Sketch look, but a nice variation for a starter Selfie image. See small image below for the variation. For this image Kyle’s Real Watercolor Detail brush set to Normal brush mode was used a lot. A Mixer brush was used to smooth the face somewhat and the hand had to be reconstructed using the Detail brush and the SJ 36% Blend for face from Grut’s Late Never brush (for settings see my A Few New Selfie AI Tricks blog at bottom – this brush is indispensable for me and the Selfie images). Then the image was opened into Anthropics Portrait Pro. I had never used this plug-in before, but it really gives some interesting results. My Selfie looks like she came right out of the Barbie Doll Collection??? Plan to do a review when I can. Really a lot of fun to use but probably does not work with most artistic media Selfie faces!
For the image above, instead of using Dave Kelly’s action, Unmesh Dinda’s Technique was used – see his Shorts Video Turn Photo to Painting with Generative Fill in Photoshop on YouTube. (If you understand the HSB Brightness concept, he changes the B amount set to 43.) The image above was a bit of a surprise to me when it showed up as a watercolor variation of the Selfie image. In this case, only the Selfie original subject was selected using the Quick Selection Tool to remove the background before applying the Generative Fill command. You can see this in the original variation shown below.
It took several Generative Fill iterations – this time just the prompt “Watercolor” was used for this portrait image. The background was selected, and instead of creating a Firefly AI background, the watercolor trees and flower were from the Let’s Travel Watercolor Set. The Watercolor brushes used on this image were Kyle’s Real Opaque Square, Diluted 90, Watercolor Eraser, and Wamazing-Basic – all in his watercolor set that is free with Photoshop (download from Brushes Panel Hamburger icon and select Get More Brushes).
******
This is another variation from the watercolor generated group and used the child image’s process. The background was changed somewhat. See below for original variation result. A second Generative Fill layer was used to replace the squinty eyes (used prompt watercolor eyeballs). Only one eye in the variations looked good, so it was duplicated and flipped, painted over somewhat, and the eye reflection was clone stamped back in (and the Blur Tool was used on the eyes to soften the lash edges). My favorite brush for this image was Kyle’s Real Watercolor Stain Damp Paper 3 which was used to clean up the background on her coat. When doing a special type of Medium, in this case watercolor, I make copies of all the brushes being used for the Selfies so they are all together to use on other Selfies – and often I save the brush to the size I like and not the default size of the brush. The fonts, which work together, were Style Casual and Style Endings with a colorful Pattern Overlay Layer Style over it.
******
The image above used Dave Kelly’s Painting 40 action on the Selfie (Quick Mask needs to be set to Selected Area) – then in the Generative Fill command prompt, typed “Cubist Portrait.” This did not turn out quite like a cubist image, but overall it has an interesting quality to it, so it made a pretty pleasing Selfie effect. The hair was created using Kyle’s Paintbox-Bristle Supreme from his Megabox. The dark lines were created using the Staccato Ink brush from Kyle’s Winter 2024 set. Below is the original variation used to start this creation. Not as much was done on this image as on some of my others. This image mainly used these brushes: a 4-pixel 80% Hard Round brush was used for all of the glasses – this brush is so handy since it gets into tiny areas easily. If made a little larger, it can be used to give really sharp edges like lot of the Cubist images have. A regular Chalk Brush was used to make the background more solid.
Below is a slideshow of several of my recent Selfies that were posted up on my Tidbits Blog (click at top of blog for link where more info is provided on each) and on Flickr in case you missed them.
As a review, once I find a variation to use, I create a New Layer above and go to Image -> Apply Image to place the variation on its own layer. To put this layer in a New Document, right click and select Duplicate Layer and in the dialog drop-down menu, change the Destination Document to New. I save it now and use this file for doing changes and painting to the variation. (The default is Merged. If just a simple change was done, like some eyes created, they can be put it on its own layer by creating the new layer and selecting the “eyes” AI layer in the drop-down, not New – just the eyes will be on the new layer in the original file.) Still have all the other variations in my original file in case I want to use them later or want to generate more variations. The ones that are bad are always deleted.
These are really fun to do and it is a great way to learn how to use all the great Photoshop brushes available, even just using Kyle’s that come with Photoshop. If you have any questions on how to do this, feel free to leave a comment. I am always trying to improve on the process. As far as the AI goes, I am still a bit “on-the-fence” about it, but by using it, you start to learn the power it has. Have a good day!…..Digital Lady Syd
Digital Lady Syd here wishing everyone a very Merry Christmas and Happy New Year! Time to enjoy yourself before getting back to business! My AI selfie above seems to be a bit on the serious side, but I complement her on wearing a lovely green jacket for the season!
Won’t go into a lot of detail with this blog. If you want to see the process being used, just check out several of my last blog posts or go to my Tidbits Blog (click at top of blog to link to it) for more on the brushes and some tips used. Still a learning process but it has been fun! Give me a comment if you have any questions on how to do this.
I will just add a couple things. The background was actually created in Adobe Firefly (free to us Photoshop Creative Cloud users). It gave this really wintry scene (prompt used was Fir tree with snow outside) – just brought into PS as a jpg to use for the background layer after selecting the subject from the original car interior background. I do believe you get different AI looks in Firefly. The subject was generated using the prompt “digital oil painting” for a change – the paint strokes are not quite as heavy as what the “oil painting” prompt gives. Painted her face with Kyle T. Webster’s new 2024 Winter Set Thinner Oil brush (download from the Brushes Panel’s upper right hamburger icon and select Get More Brushes – new set at top of his page). Also used Kyle’s Great Paint brush from his 2023 Fall Set (at end of Kyle’s brush download page).
I am wondering what my selfie will be doing for 2024!…..Digital Lady Syd
For a few weeks I have been working hard on getting an AI generated effect I like. Still using my same selfie image from previous selfie posts which has helped me figure out what works and what does not on AI portrait images. All of the selfie variations were created by using a prompt “Oil Painting,” but some had more specific information added to the field. This image uses the same brushes (I relisted them at end if you want to see them along with a new Smudge Brush.) Below are the steps that work really well to get some interesting portraits to paint.
Steps for Creating Portraits From Selfies:
Open selfie. If you want the person to be generated, after you run Dave Kelly’s free Gen Fill Photo Painting Action (usually set to GF Painting 40 for the effect shown in my selfie images – really need to use Dave’s action as it is so easy to use and be sure to try his different Brightness amounts), the selection is the whole image.
To get a good portrait result, need to either go to Select -> Subject and OK to discard your current selection, or use the Quick Selection Tool or the Lasso Tool to subtract from the selection to get just the subject selected. Run Generative Fill and look at your variations. Can Generate as many as you want.
For creating a background, in the Generative Fill taskbar select Subject and Invert Selection – in prompt field type what text for type of background wanted. Can change the prompt to read anything and use any type of style. The prompt used here was “oil paint snow scene.”
Create another Apply Image on a New Layer placed above this layer to create one with these variations merged into one layer.
Turn off all but the top layer so the brushes used to clean up the image will work faster. On another New Layer above just start painting and have some creative fun!
Some Selfie Tips
If you want to use Generative Fill to change something like her lips, which were changed from the original variation on the above, select around the lips with the Lasso Tool and run Generative Fill – the prompt used here was “oil painting lips” and three variations appeared. Add a New Layer above for another Apply Image layer, but this time instead of using the default Merged Layer, open the drop-down and select the Generative Fill lips layer. The lips will now appear on its own layer and can easily be manipulated. In this case the lips were reduced a little in size, and a layer mask was applied so just part of the new lips were added into the image. Be sure to turn off the Generative Fill lip layer to work on the Apply Image lip layer. Used paintbrushes on another New Layer above the lips to blend everything together.
Another really great tip for cleaning up these AI images, which usually have strange things going on in certain areas like the face, is to use Photoshop’s Liquify filter. This image used the Face-Aware Liquify section for the jawline to make it a little less harsh and on the nose, and the Pucker Tool to adjust a little from her ear. This is so incredible!
It is easy to add jewelry to your subject – just make sure you have a Lasso selection exactly where you want the jewelry to be added. I had trouble getting her necklace to look straight on this image – AI kept making it squiggly. Finally got a selection to fit properly. Used in prompt “oil painting diamond pendant,” and for earring used “Oil Painting Earring” – neither is exactly what I had in mind, but they look pretty good.
One last thing, 5 AI Generations were run on this image – background, lips, nose, pendant, and earrings. In all cases, an Apply Image was created (just highlight right above the generative fill layer with the correct variation showing), and the original AI generated layers were placed in a turned off group (might want to use a different variation later or rerun the Apply Image if you really mess up the image). Only used the Apply Image layers as the working layers.
When painting, be sure to go between the regular paint brushes, Mixers and Blender brushes, and Smudge brushes. I am constantly going back and forth. I have my Mixer brush set to A and Smudge Tool set to U in the Keyboard Shortcuts so I can flip between them quickly. And if a brush does not look right, change some of the settings like the spacing or the Angle Jitter Control or Scatter Amount. If you like the brush, just create a brush preset in the Brush Panel.
To finish off the image, just used my regular photo workflow: Black & White Adjustment Layer, Color Lookup Adjustment Table using 3D Lut File FoggyNight from PS at 55% layer opacity, and a Curves Adjustment Layer. The Snow Overlay is from a free set I created at deviantArt – PNG files so no blend mode needs to be used, just adjust the layer opacity (50% here) and use a Layer Mask to remove snow off certain areas with a brush.
All these images were created from my same selfie image and I keep going back to it to run more variations. Once you find a good selfie to use, it seems logical to use it as the template for other images. Hope this blog made this seem a little easier – it is really fun to do!…..Digital Lady Syd
Painting Brushes I am Using for these Selfies.
Using mainly five brushes:
I like Kyle’s new Fall 2023 brush Great Brush (changed Flow to 54% and Smoothing to 12% in the Options Bar) using a small size for adding clumps of color.
Grut’s Watercolor Late Never brush modified to work as a regular brush blender – works fantastic (here are the settings if you want them: I call it “SJ – 36% Blender for face-from Grut-W Late Never.” set brush in the Options Bar to Size 35 pixels, Mode Normal, Opacity 36%, Flow 100%, Airbrush icon off which is the Buildup in Brush Settings Panel. Then in Brush Settings Panel set Shape Dynamics Size Jitter 0; and uncheck Buildup (should be off if Airbrush icon off in Options Bar) and Protect Texture (not sure why this is even on). I have several variations of this brush, but this one is best for blending, especially with the Great Brush – just sample and blend between colors to smooth them out.
Fay Sirkis Mixer brush called Short Streaky Detail Blender (not sure it is still available anymore) – any small detail brush mixer would probably work to get the hair look.
Smudge Brush was created using Kyle’s All Purpose Blend Smudge Brush and turning off the Spacing in the Brush Tip Space Brush Settings Panel, Shape Dynamics (no Size Jitter and Control for Size Jitter), Angle Jitter set to 3%, and Angle Jitter Control set to Direction; Transfer checked with 0 Strength and no Control; Noise checked, and in Options Bar Strength set to 92%, and Sample All Layer (check or not) – I named this brush SJ-Painterly Smudge Brush and used it on the background. The cool thing about this brush is that if the ALT key is held down when starting to blend, the Foreground Color appears – this is great for adding in a little color to accentuate the blending. Used this brush to blend some of the harsh lines created in the background on the above.
Just wanted to blog out another pretty cool Photoshop AI trick so that you can work on numerous variations at the same time in the same or in different files. Learned this from Aaron Nace, another really great Photoshop guru that always has lot of tips and tricks to share.
Once you have run the Generative Fill command in Photoshop, there is a way to add each variation onto its own layer and it very simple.
Highlight the variation you want to place on its own layer.
Create a New Layer above the Generative Fill AI layer and highlight it.
Go to Image -> Apply Image. Just use the default settings. Okay.
Now the variation is on its own layer. As long as you have the variation you want on its own layer showing up in the Properties Panel, an image can be made quickly and you can make several layers from the same Generative Fill variations.
This can be very useful, especially if you want to combine two variations into one image. Sometimes you like a sky area in one variation but not the foreground for example. Just run this for each variation and you will find a new layer for each variation in your Layers Panel. Just add a layer mask to one and remove what you do not want to see. And you can put your main subject into different backgrounds by selecting the subject and inverting it – then run all kinds of backgrounds and having different files to put in a screenshow. That is what Aaron did.
The girl above is another variation from my selfie image (Dave Kelly’s free Gen Fill Photo Painting Action set to GF Painting 40 and Generative Fill Prompt: Oil Painting). Put her on her own layer – then created a New Document to do my painting and adjustments to the image. Used four brushes – two Regular brushes: I like Kyle’s new Fall 2023 brush Great Brush (changed Flow to 54% and Smoothing to 12% in the Options Bar) using a small size for adding clumps of color, and Grut’s Watercolor Late Never brush adjusted for smoothing like on her face (for settings, check the Brushes Used on These Images at the end of my Creating an AI Selfie? blog – also more variations from this same selfie image).; and two Mixer Blenders: still using David Belliveau’s free Mixer blender at 15 px, and an older Fay Sirkis Mixer brush called Short Streaky Detail Blender (not sure it is available any more) – any small detail brush mixer would probably work to get the hair look. It is important that you find a few brushes to use for painting clean up on these AI files – they generally are pretty rough.
Well, this was short and sweet but I thought it might be useful to you when trying to figure out how to be creative with this new AI capability. …….Digital Lady Syd
Just had some fun this week – thought I would try Photoshop AI’s Generative Fill on an old selfie pix taken inside my car while it was raining really hard. I also took some texture images of my moon roof with the water drops on it. Dave Kelly’s Gen Fill Photo Action Painting using GF Painting 40 was applied to all the images and then the Contextual Task Bar Generative Fill command using the prompt “Oil Painting” was applied on the first three images and “Cubist Portrait” on the last one. The GF Painting 40 action means that 40% of the image is rather abstract while the GF Painting 80 action gives a more realistic effect. I think using this action is the easiest way to get good results when using the Generative Fill Command. (To get links for action and info about it, go to my short Tidbits Blog Older Woman of Another Generation.) Got lots of variations – some good and some not and sometimes my white coat showed up in a variation. For the image above called Raining Tears, the raindrops appeared on the subject and her coat which I thought was pretty cool!
For the above which may be the closest to my actual selfie (well she is a little younger but she has my coat on!), the subject was selected by removing part of the GF Painting 40 action selection (note Dave’s action selects the whole image for setting the brightness in Quick Mask Mode), so the Quick Selection Tool with the ALT key pressed was used to remove the background area from the prompt requirement. With the selection still active, the prompt “Oil Painting” was entered in the field. The interior of my car from the photo was left intact while the subject was oil painted. Felt the background was too realistic, so turned off the subject AI layer, highlighted the original selfie layer, and ran the action again applying GF Painting 80 – then ran the Generative Fill command with the “Oil Painting” prompt. Chose a variation and just turned on the top original AI layer – now had two different strengths of oil painting in the same image. BTW, used an oldie but goodie filter, Topaz Lens Effects applying the Gold Left Reflector preset and the Plush Rose Vignette preset at a 12% opacity – gave a nice warm feel to the image. The brush for clean up was my SJ New Chalk Brush (see settings below).
This image was actually a different variation from the top Raining Teardrops image. Only this time the background was selected, just like it was done in the second image, but cleared so a texture could be added. This time a texture was made using the Stamp Tool for the background. (See end of blog on how to do this.) Used all brushes but the chalk one listed at end. The ear lobe was selected to create an earring using Generative Fill with the prompt saying “Oil Paint Earring.” Otherwise the same workflow.
Played action Dave Kelly’s GF Paint 40 and used Generation Fill prompt “Cubist Portrait.” Ran several times but ended up with this variation created in the second generation effort. It did pick up my white jacket also. An Impasto Layer was used to add a touch of texture in the paint strokes on a separate layer and a scatter brush was used to paint over it. (To learn how to do this, check out my older but still good Fun Photoshop Blog and video called How to Create an Impasto Layer Style.) Also painted with a flat square brush areas that had too much impasto on a separate layer above.
As promised, keeping it short. It was fun to try this just to see what it does. As you can see, they are not exactly your image unless you use a very high amount with Dave’s action. It was kind of a crazy but fun thing to do though! Have a good one!…..Digital Lady Syd
ADDITIONAL INFO
Brushes used on these images: And there was a lot of digital painting done using a couple major brushes – Kyle’s Great Brush from his Fall 2023 set (this brush is really great so download it from his “Get More Brushes” link in the Brush Settings hamburger menu); and Grut’s Watercolor Late Never brush that I modified to work as a regular brush blender – works fantastic (Here are the settings if you want them: I call it “SJ – 36% Blender for face-from Grut-W Late Never.” Changed Grut’s Late Never brush in the Options Bar to Size 35 pixels, Mode Normal, Opacity 36%, Flow 100%, Airbrush icon off which is the Buildup in Brush Settings Panel. Then in Brush Settings Panel set Shape Dynamics Size Jitter 0; and uncheck Buildup (should be off if Airbrush icon off in Options Bar) and Protect Texture (not sure why this is even on). I have several variations of this brush, but this one is best for blending, especially with the Great Brush – just sample and blend between colors to smooth them out. Second image brush was just a the Legacy Default Photoshop 36 pixel Chalk brush with a some extra settings: Brush Tip Shape – Angle 15 degrees; Shape Dynamics – on a variation the Angle Jitter Pen Pressure is set to Direction but the main brush has no settings here; Texture – pattern called Pretty Actions Antique but it is basically a medium gray and white marbled look, Scale 87%, Brightness -35, Contrast 75, Texture Each Tip checked, Mode Multiply, Depth 70%, Minimum Depth 63%, Depth Jitter 90%, and Control Pen Pressure; Dual Brush set to Color Burn, Brush Tip oval shaped, Size 200 px, Spacing 29%, Scatter 11%, and Count 7; and Transfer – Opacity Jitter 100%, Control set to Pen Pressure, Minimum 0%, Flow Jitter 0%, Control set to Pen Pressure, and Minimum 0%. (This is a really nice painting brush.) For third image Kyle’s Munch Medium brush was used on the face. The Edvard Munch seven brush set first appeared in 2017 in a 4-part video series called Get Started with Digital Painting Photoshop – they are not part of the PS2021 brushes, but are a free download here at the Adobe Creative Cloud.
Creating the Pattern for behind the 2nd image: First added a new layer, selected the Pattern Stamp Tool and used a brush I created a long time ago called SJ FrenchKiss-Tableau Mirage. (See my older Fun Photoshop Blog How to Paint with a Texture Brush from Your Image with video to show you how to create a brush from any texture.) Obviously French Kiss‘s Tableau Mirage was the texture converted to a brush and made large – 3000 px. I use it all the time for this type of background. On this image Jessica Johnson‘s The Masters Palette Pattern TM103 was selected and just stamped once in blue (note it is not totally opaque) – it can be placed in different places and select different colors for the texture wanted. (Jessica is definitely the Pattern Stamp Queen of Photoshop and has some great ways to using it.) Next created another New Layer and used the same brush with a brownish color and selected a stamp pattern in same set called TM142 – this causes some of the blue to show show through from TM103 but uses the main color light beige color from TM142. Don’t have to use the Pattern Stamp Tool for this.
Decided to write this blog just because it has been so much fun creating these dogs and I wanted to share how to do this. I think most people will be able to get some nice results very easily even if you do not do much drawing. Kyle T. Webster’s YouTube video called Digital Drawing Workout: Drawing Dogs with Custom Brushes was used as a guide for creating all these dog images. The brushes in the video are all ones Kyle created – I copied each brush into a Photoshop Brush Dog Group in the Brushes Panel so they could be found quickly while drawing. Also added other relevant brushes (see end of blog for more on this) and some variations of Kyles’s brushes that helped create some of the dog differences.
These dogs all followed the steps Kyle discussed in his video. The first step is to find an image of the breed of dog to use as a reference for starting your drawing. All the blog images contain a rough drawing layer that used his Megapack->Inkbox->Kyle’s Inkbox-Brush Beauty brush set to 25% Smoothing in the Options Bar for sketching. (If you do not know where to find Kyle’s free brushes for Photoshop (and Fresco), open the Brushes Panel and click on the upper right hamburger icon drop-down menu – go to Get More Brushes – log in if needed. There you will find for free all of his over 1000 brushes in different sets for download.) Brush Beauty actually worked pretty good so it was used as the sketch brush for all my images (saved it as a preset brush variation). After sketch layer is finished, the Lasso Tool was used to select the whole dog – could use any accurate form of selection. The dominant color of the dog, as shown in my rendition of his demo Greyhound dog below which used white, was painted inside the selection using his Spring 2022 set’s Chef Maltese brush – it adds great texture (and Kyle says to adjust the Brush Settings Panel Color Dynamics Brightness Jitter slider for even more texture)! For the other paint and detail layers, be sure to clip New Layer (ALT+click between your New Layer and the solid paint-filled dog layer) so paint strokes do not go outside the edges when adding the new paint colors. The Woodchop Joey brush (really like this brush for the dog coats) was used to paint in the darker tones and pinks in the Greyhound below. The brush was set to 50 pixels (and saved as a brush variation preset) and used as a detail brush on some areas. Woodchop Joey was also used on the Bulldog above. The shadow areas on the Bulldog used the Megapack->Kyle’s Drawing Box->Kyle’s Drawing Box- Graphite-Control 2 brush with tones of gray. The Spring 2022 set’s Suavy Inky brush was selected for both of these dogs to paint in the dark details. The Beagle used the Chef Maltese brush for painting the dog selection in white, Kyle’s Real Watercolor set’s Warmazing Extra Rough brush for the color areas, and Suavy Inky for details, .
As you can see, pretty much the same brushes for these smoother haired dogs. All the images are shown without the original rough draft layer showing. Had to put up an image of my darling Springer Spaniel I had for many years. Pretty much the same process – white used to paint in the dog selection using Chef Maltese brush, and the dark areas used the Woodchop Joey brush at both large and small sizes. Kyle’s Megapack->Paintbrush->Kyle’s Paintbrush-Bristly Fat Flex was used on her face and a small dotted hair brush to get the hairy edges on the dog coat.
Kyle made a white poodle in his video. He used a brush from his Spring 2020 set called the Bouquet 1 brush which created the great big puffs of hair, and by sizing the brush down gives the smaller puffs. My Poodle used Woodchop Joey for the legs and some of the browner colors in the face. Also his Megapack->Drawing Box->Kyle’s Drawing Box-Pastella Brush was used to fill out the hair.
For the Husky Dog the brushes he suggested using were Kyle’s Paintbox-Bristly Fat Flex brush and Megapack->Drawing Box->Kyle’s Drawing Box-Conte Crayon – he did not get this dog done but I finished mine. The drawing below is my rendition of Kyle’s Sheep Dog that followed his video instructions – not exactly original but I liked the way he turned out. Kyle used a really odd brush called Geobot 1 from the Spring 2022 set – all drawn on one layer (over the dog selection, lock the transparency of the layer (the \ key or press the first icon next to Lock in the Layers Panel) and paint with this brush, just sized down the brush for the face area. Can turn on and off the transparency of layer to get the correct edges. If using a tablet, use lighter pressure along the edges and try using different sizes on different parts of the dog.
For more brush choices try downloading from DeviantArt this free set of brushes by Coyotemange called Wildlife Texture Brushes – they are great for painting in missing areas of fur on all kinds of animals and birds. By making them smaller, nice sharp edges can be made. I have used these brushes for all kinds of animal images – by adjusting the color dynamics, size, angle, etc., almost every kind of fur can be created. And be sure to save any variations you like as new brushes by clicking on the plus icon at bottom of Brushes Panel. Also, when finished with your drawings, save the Dog Group as a set by highlighting your dog brushes in the Brushes Panel; then go up the hamburger icon drop-down menu again and choose Export Selected Brushes option to copy the Dog Group brushes as a set to your computer in case you accidentally delete it.
Well I hope you enjoyed this blog – it was so much fun to do and pretty easy. Give it a try even if you are not the greatest drawer. Especially fun if you like dogs which I do. It felt good to write about something different from the Photoshop Generative Fill AI feature! Have a good week!…..Digital Lady Syd
This beautiful Mallard Duck is the major patrol guy at Ames Park in Ormond Beach, Florida. He is totally responsible for watching the goings-on at several lily ponds – very busy! Had some fun with this one as the image was actually taken on a sunny bright morning. Thought I would share how to change an image into night without a lot of work.
The first step was to get the dark but natural effect. In Lightroom (or ACR) to get the evening effect with Serge Ramelli’s American Night Darker preset (preset is from 2019 and has been called the “Harry Potter Effect.” It can be created by following his video called Lightroom Tutorial Dark Moody – Day to Night “American Night” It was used to set some lighting effects on the duck and lily flowers in the background. That is all that was done in Lightroom.
In Photoshop a New Layer was made to add in the glow effect to the duck and water shadows. Etherington Brothers‘s info sheet called How to Think When You Draw – Atmospheric Glow Tip by Steve Stark (not sure how to link but it was in their X Feed of September 13th) was used to paint in more glow on the duck and the water shadows. This involved setting a New layer to the Color Dodge blend mode. Select a soft 20% opacity brush for painting with a yellow color where the glow should be added – then open the Layer Style and uncheck Transparency Shapes Layer. Voila! This creates a really nice soft glow that can be reduced by just changing the layer opacity.
Applied a Linear Gradient Fill Layer using the 18- washed_away_art preset (a brownish to gold gradient from 2015) set to Soft Light blend mode at 49% layer opacity. Used Blake Rudis’s High Pass Sharpening Action and painted in the layer mask on just the duck mainly (see his video for getting the best detail out of a photo – it is called Don’t Use Clarity – Use This Instead). A Color Lookup Adjustment Layer using an On1-Heat Wave LUT (yes, they can be used in PS) was applied. Viveza II (still my favorite filter) was opened to adjust the color tones – new version of is linked here. Last step was a clean up layer.
This image taken at the Hilton Waikoloa Village on the Big Island in Hawaii is another example of creating a night effect from the day. Not sure this sculpture is one I particular love, but he definitely stands out when walking a little path around one of the hotels. The same workflow as for the duck was used with a few additions. A Color Balance Adjustment Layer was added to make the sky appear more like evening – mask used to just target that area. Also at the end two spotlight layers for lightening and darkening the image were used (see my blog called How to Add a Spot of Light).
Hope you get a chance to try a few of these techniques. I really like Blake Rudis’s High Pass action – it is updated from his first one from several years ago. Also the Atmospheric Glow trick is really nice and gives a nice quick pop to objects in an image. Have a good one!…..Digital Lady Syd
While watching all kinds of sports shows today, I decided to try drawing a fly using an info sheet called How to Think When You Draw – Flies – Part A by the Etherington Brothers (there is also a Part B). If you want to hone your drawing skills, these guys are the ones to help. They have info sheets on all kinds of things. It was fun to just draw something I had never tried before.
To began a rough and final draft layers were drawn of just the foreground fly using Kyle’s Animator Pencil 2016 in his Megapack Drawing Box. On a new layer under the final draft the color was added – used Kyle T. Webster’s Chef Maltese brush in the Spring 2022 set for all of the fly.
Below the fly the sky and ground were painted in using Darbuka Painter from Kyle’s Turkey Syria set (this was a set he sold for the Covid Relief effort in India). Used Aaron Blaise’s free Cloud Brush for the clouds.
Drew a leaf under the fly using the Animator Pen and colored it in with Woodchop Joey from Kyle’s Spring 2022 set also. A Drop Shadow Layer Style was selected to add some depth to the leaf (brownish sampled color, Opacity 41%, Angle 24 degrees, Distance 42 px, Spread 29%, and Size 144 px). This is all pretty standard drawing stuff!
Then I decided to add some more flies into using Photoshop’s AI so a stamped layer was created on top (CTRL+ALT+SHIFT+E) and duplicated (CTRL+J). Using the Lasso Tool, the sky and side areas were selected to indicated where the flies should go while being careful not to include my own drawn fly. In the Generate Prompt box “Small drawn cartoon flies” was typed. After generating, three more versions of my own fly appeared (see screenshot below as an example) – AI used my own drawing as the basis for the additional AI flies in the image! I was totally surprised! Apparently it will take your own design and add it back in. A second Generate was done and different configurations of my flies appeared! The image was taken into the Photoshop Beta just to see if this was a fluke and it also generated more of my flies! Thought this might be something you would find fun to try since it does not seem obvious it will do this, and it definitely can blend the new subjects into your design nicely.
I did not like the extra parts that were added into the image, so this time I duplicated the original AI layer and turned it off in case I messed up the image. On the copy where I wanted to apply the Layer Mask, it is grayed out. To make it work, the layer has to be rasterized first and then the Layer Mask can be applied. The Lasso Tool was used to select around the new flies and inverted so the background could be removed using CTRL + Backspace. Used the Eraser Tool to clean up round edges from the selection. A new layer mask was added and the Gradient Tool was applied to soften the flies in the background. A Color Lookup Table Adjustment Layer (Photo Focus-Joel Grimes Indian Summer preset from an older Luminar Set), a Black and White Adjustment Layer set to Luminosity blend mode (see my How to Use a Black & White Adjustment Layer to See Contrast in an Image blog), and a Curves Adjustment Layer were the final steps (see Denny Tangs Tone Chart Photoshop Action – an oldie but goodie but has made a huge difference in my images for years). The text was called Modernline.
Hope you enjoyed this little AI discovery I made – sort of not what I expected! And I hope you are giving the new AI feature a try – it is a lot of fun!…..Digital Lady Syd
Digital Lady Syd's Fun Photoshop Blog 