How to remove exposure blending halos

One of the biggest frustrations I hear when photographers try to use exposure blending with luminosity masks to get beautiful skies is that they frequently get halos around hard edges such as the point where trees or mountains touch the sky. In my Exposure Blending Master Course, I teach several techniques on how to avoid halos in the first place. However, there are many times when you may still encounter halos for a variety of reasons, and it is important to know how to remove those halos easily.

With Lumenzia v8, you can now use the “Edge” button to do just that. It will quickly help you select the edge of any mask or selection so that you can paint white or black on your mask to remove the halos. See the video and the written summary below to learn how to save time and cleanly edges in your exposure blends.

 

 

Here’s a summary of the workflow to remove halos from the luminosity mask using Lumenzia:

  1. First, identify the cause of the halo.
    • Are you seeing too much of the dark layer on top? Too much of the bright layer below? It’s critical to understand exactly what the issue is in order to know what fix you’ll need. Toggling the layers and masks off and on is an easy way to check.
  2. Once you know the problem, identify the solution you will need to do to the luminosity mask to remove the halo.
    • Do you need to paint more white to reveal the dark sky layer, or black to remove some of the sky?
  3. Next, <ctrl/cmd>-click the luminosity mask to load it as a selection.
    • The marching ants will show you roughly where the current edge of your mask is. Remember that the marching ants do not show pixels darker than 50% gray in the mask.
    • Compare the rough edge of the selection to the solution you identified in the previous step. Do you need to select pixels which are inside or outside the existing edge? By how much?
  4. Use the “Edge” button in Lumenzia to convert the selection of the masked area into a selection of just the edge where you need to paint black or white.
    • In the left side of the popup dialog, make sure “edges” is selected.
    • Select a radius of about 1 or 2 pixels. It’s typically best to try starting with just 1 pixel, as painting outside the edge of the existing halo will likely just create a new halo.
    • Select the number of pixels to expand (positive numbers) or contract (negative number) based on your evaluation from the previous step. Something between -2 and +2 is typically ideal. Due to the nature of this tool (with the radius and feathering), the edges tend to grow a little from where you think they are to begin. You also want to err on the safe side (so the center of the new selection shouldn’t go right over the halo, or the edges of the selection will go beyond it). Therefore, you will likely need to subtract about two pixels from what you would otherwise expect. For example, in the demo above would probably have guessed +1 pixel, but the correct choice was -1 to properly target the edge. So be sure to experiment with things a bit to find the best settings for your image.
    • Click “Selection” to create the edge selection.
    • You will be prompted to choose how much to blur the selection. The default generally works well, but you might try up to 1 or 2 pixels if you aren’t getting the results you like in the following steps.
  5. Given the work you’ve done to create this selection, now is a good time to click “Sel” and save the selection. If you need to create a couple different edge selections, this will be very helpful to switch back and forth.
  6. Now paint black or white as needed onto the layer mask through the edge selection you have created.
    • If you aren’t getting the desired results in general, deselect and start over from step 3.
    • If the selection works in some areas, but not others, repeat steps 3-5 as needed to create multiple different edge selections. If you have a complicated sky or foreground where the tonality changes a lot, you will likely need to use different selections for different areas of the halo.
  7. If you find a few pixels that stray outside your selection, don’t be afraid to paint free hand. This is quick and easy to do for small problems in the image. Just deselect (<ctrl/cmd>-D) and re-select (<shift><ctrl/cmd>-D) as needed to temporarily paint without the selection.

 

How to Edit a Bright Building against a Bright Sky

When your subject has similar luminosity and color to its surroundings, simple luminosity masks often struggle to get the job done. Neil Williams ran into this scenario when he shot the image below. He emailed me asking how he could use Lumenzia to help brighten this white temple against this relatively bright sky. This is a great example of a common problem, and he offered to let me use him image for this tutorial.

Luminosity masks are well suited to a challenge like this because the building is slightly darker than the sky, and alternatives like manual brushing or the pen tool would be a nightmare with all of the fine detail along the temple and trees. However, no single luminosity mask will adequately separate the temple from both the sky and trees. So we need to use a more advanced combination of custom luminosity masks and some hand painted layer masks to get the job done.

Here’s a summary of the workflow used to edit the image:

  1. Remove dust spots and distractions. This should ideally be done first, to avoid including dust spots in the luminosity masks
    • Click “✓dust” in Lumenzia to visualize the dust, as well as create a blank layer for cloning and activate the spot healing brush.
    • Paint with the spot healing brush to remove the dust spots.
    • Click “✓dust” again to cycle through the various visualizations to check for anything that was missed.
    • Click “X” to clear the dust visualization.
    • Use the spot healing brush to remove the lamp.
  2. Lighten the building. The masks involved here are the trickiest part of the edit.
    • Create a new levels adjustment layer. Because the brighter sky makes the histogram useless for determining how much you can safely bring in the white point to edit the building, a good way to safely brighten the building is to hold <alt/option> while bringing in the white point and watch to see when the temple starts to clip and then back off a little. Note that waiting to adjust levels until after you add the layer/luminosity masks won’t improve the histogram, as it is based on the layers below and ignores the masking on the levels layer (you could mask the underlying image, but then the transparent pixels show as lots of white in the histogram).
    • Temporarily hide the adjustment layer (so that it does not affect the luminosity mask in the next step).
    • Add a midtones mask by clicking “(c)” and then “Mask”. This mask helps achieve a few goals. It avoids lightening the shadows of the temple, which creates better contrast when lightening. It also minimizes any targeting of the dark trees or bright sky (it does not eliminate them on its own, but is a simple way to help reduce targeting of the sky).
    • Make the adjustment layer visible again to be able to see its impact.
    • Add a group with a black layer mask. This allows us to selectively paint white to reveal the adjustment where needed. A simple soft, white brush is all that is needed here. The adjustment will spill into the sky and cause unwanted lightening, but we’ll fix that in the next steps.
    • Add another group, but with a white layer mask this time. We will use this to paint out the sky, which should not be lightened. This is the most important mask for separating the building from the sky.
    • Temporarily hide the adjustment again while creating any luminosity selections (as the building is better separated before lightening it). Create sky selections as needed (you’ll frequently need more than one to paint out all critical sky areas).
    • I clicked (e), adjusted to 9.99 gamma, and clicked “Sel” to make the first sky selection. The gamma adjustment helps create a stronger selection, without clipping at the sky edges. Paint with a soft, black brush on the white group mask to stop brightening the sky. You should have the levels layer visible while brushing to see what effect the mask is having.
    • I then clicked L3, adjusted white levels, and clicked “Sel” to make a secondary sky selection to be able to paint black on a few pixels that weren’t included in the first sky selection. I choose to use the whites slider in the levels adjustment here because the mask was starting from a darker point and didn’t have as much risk of clipping.
    • It is frequently helpful to use “Sel” to save your luminosity selections when working advanced/custom selections, so that you can reload and reuse them later if you find any areas that need adjustment in the mask.
  3. Enhance sky color.
    • Create a new HSL adjustment layer
    • Set the targeting to red, set saturation to +51, and drag out the blue slider to include the entire sky with a natural look.
  4. Keep the building relatively neutral. The sky enhancement has also affected the building. Rather than going through the trouble of creating a new sky selection, we can simply adjust the saturation of the building through the masks we have already created.
    • Create a new HSL adjustment layer right above the levels adjustment. This is inside the two groups and therefore has nearly the same masking as the levels layer. In this case, the midtone Z(c) mask isn’t needed for color adjustment, but you could also create another group to put both adjustment in and this would give you the exact same mask for HSL that we already created for Levels.
    • Set saturation to -45.
  5. Crop the image as needed. I like to do this late in the process to keep options open to change the crop later if needed, such as to optimize for a specific print size.
  6. Add a vignette by using the lasso selection tool and clicking “Vignette”. I should have ideally done this before cropping.

 

Here are some additional optional steps you might need to use in your own images if you run halos or some dark spots that should have been lightened:

  • If you find halos (bright pixels at the edge), you can quickly repair those with the “Edge” tool in Lumenzia v8+. This is particularly helpful for eliminating small issues which may become visible in large prints. The key to fixing them is to expand the area blacked out of the sky. The steps include**:
    • <cmd/ctrl>-click the top group mask layer mask (“”Group 2”, the one blacking out the sky). This loads it as a selection.
    • Click “Edge” in Lumenzia and choose to isolate edges with a radius of 1 and -1 to contract by 1 pixel. This selects the halo area by pulling in from the existing edges.
    • Click on the “Group 2” layer mask to target it and paint with black through the halo selection you just created. This paints out the edge pixels to remove the halo.
  • If you find dark spots in the building, this indicates that the top group mask (“Group 2”) blacked out some areas that were not sky. Just paint white free-hand on the that “group 2” mask to fill in any gaps. No luminosity selections are needed to safely paint this isolated interior problems in the mask.

 

I’ll try to add a demo of this use of the “Edge” button in the future. If you are in my Dodging & Burning Master Course, you can see such an example in section 9.3 video 9h where I use “Edge” to remove some halos created while dodging.

How to optimize your “gray working space” for better luminosity masks

Few aspects of photography cause more confusion and frustration than “color management”. I wouldn’t blame you for wanting to throw your hands up and just pray for decent results rather than trying to figure it out. But it’s worth investing some time in understanding it if you want the best quality images.

It turns out that color management affects luminosity masking in surprising ways. If you optimize properly, you’ll get the luminosity masks and selections you expect. If you don’t, you may find issues such as white in areas of your layer mask you thought you protected. (Note: If you believe you’ve got this covered by setting your gray gamma based on your RGB working space, be sure to keep reading… that setting does not have the impact you may be thinking).

There are several Photoshop preferences for “working spaces” (under Edit / Color Settings). These are the color spaces Photoshop will use there is some ambiguity. For example: If you go to Image / Mode / RGB, your image must be converted to a specific RGB profile, but this menu command gives you no way to specify which one. So Photoshop will use the RGB working space in that case (unlike when you use Edit / Convert to Profile, where it will use the RGB profile you select). Or if you open an RGB image which does not have an embedded profile, Photoshop will assume your working RGB profile for this ambiguous image. But that is rare, and the RGB working space setting won’t matter if you are working with images with embedded profiles (which you probably are and definitely should be).

 

The importance of the Gray Working Space

The Photoshop preference for the “gray working space” sounds like it wouldn’t affect your color images. But it actually has a significant affect on your RGB images if you use luminosity masks and selections, as channels are always treated as grayscale (even within RGB image). The reason is because your RGB pixels (even if they are black and white RGB) must be converted to grayscale pixels when working with luminosity masks. The midtone gray values for both RGB and grayscale spaces are specified by a “tone response curve”, which is typically a “gamma” value. The math involved is complicated, but the bottom line is that your RGB “gamma” and grayscale “gamma” should match if you want high-quality luminosity masks and selections. Otherwise, the conversion will be too dark or too light (i.e. the luminosity mask will select too much or too little). And in the case of painting through luminosity selections, things get worse with multiple brush strokes, resulting in problems with exposure blending or dodging and burning.

Photoshop does not automatically match “gammas” for you (though Lumenzia will as described below). Photoshop just assumes the “gray working space” is correct and uses it no matter what RGB space is embedded in your image. And unlike the RGB space used for your image, the gray space for your masks/selections is never embedded so the grey working space always affects any conversion between RGB and grayscale. This includes the creation of luminosity masks, luminosity selections, and even the way your paint color may affect your layer mask when brushing without luminosity masks or selections. And because it is the relationship between the actual RGB space (the embedded profile, not the working RGB color space), the correct working gray space choice is not based on the working RGB, it depends on the document on which you are working. So you should ideally be confirming/updating the gray working space under Edit / Color Settings every time you open a new image or change documents if you want the highest quality luminosity masks and selections.

 

What happens if you use the wrong gray working space?

The most common gammas used are 1.8 (ProPhoto) and 2.2 (Adobe RGB is 2.2 and sRGB is very close though best matched using “sGray” in newer versions of Photoshop). If you use more exotic color spaces, you may run into other gammas (such as gamma 2.4 for Image-P3 or gamma 1.0 if you use lines profiles like ACES for video or computer graphics), and these problems will likely be worse when mismatching these more unusual gammas.

If you use the wrong working gray space (comparing gray gamma 1.8 vs 2.2):

  • The mask/selection will be stronger than expected when using the wrong gamma with ProPhoto RGB (wrong being 2.2 here). This is is bad because it means that areas you expect to be protected will likely change far too much when painting through a selection, resulting in dodging or blending areas of the image which should be protected. The difference can be substantial. For example, what should be a 3% selection would become a 5% selection, or 10% becomes 15%. So each time you paint outside the lines, the damage builds up an extra 50% or more in areas which you want protected. Completely deselected pixels are still completely deselected, but luminosity selections include of near deselected values that protect your image too. Getting the wrong results at the transitions is a recipe for frustration and poor results.
  • The mask/selection will be weaker than expected when using the wrong gamma with Adobe RGB or sRGB (wrong being 1.8 here). This isn’t ideal, but is less of a concern, as you won’t be prone to accidentally painting in areas you thought were protected.
  • If you create masks without using luminosity selections, they are also affected, but the impact is not as great. When working with selections, the error gets worse with each brush stroke and can therefore have a greater impact. So creating luminosity masks by painting through luminosity selections (which you should do for exposure blending or dodging & burning) will mean that mistakes in the gray working space are more problematic.

Of course, there are other gray spaces and they have different degrees of failure. If you never set your gray working space, it is probably some default 15-20% “Dot Gain”. 20% Dot Gain is closest to gamma 1.8, and 25% is the closest to 2.2 (30% is also close, but brighter and therefore a worse error). In any case, you shouldn’t use any of the dot gain spaces if you want optimal luminosity masks.

Lastly, you might be wondering how the choice of working gray space if you work in black and white. It does not matter for two reasons:

  1. Black and white work should be done in RGB mode, not in Grayscale mode. The only benefit I can think of for grayscale color mode is that it creates smaller documents. RGB offers the ability to use the original image color to make more targeted masks and selections, access to the full range of Photoshop and 3rd party tools (including luminosity masking software), and the ability to add a bit of color tone to your black and white image. Unless you work in newsprint, I strongly recommend RGB for your black and white work.
  2. Even if you disagree and wish to work on your in grayscale mode, you can embed the grayscale profile so that the working gray space does not affect your work (just be sure to convert to grayscale by using Edit / Convert to Profile instead of Image / Mode, so that you can specify which grayscale profile you wish to use and ignore the working gray space).

Also note that if you convert RGB profiles, your layer masks are not adapted during conversion (your grayspace does not matter, so any change in the RGB gamma will produce different layer masks if there is any other than pure black or white in them). As a result, the appearance of your image may change substantially (this is a good example why Photoshop sometimes warns you to flatten layers when making changes). This is a very good reason to create a FLATTENED copy of your ProPhoto RGB image before converting to sRGB for the web (Adobe RGB is fine because the 2.2 gamma in Adobe RGB and L* tone response curve used in sRGB are very similar).

 

What to do if you use Lumenzia?

Fortunately, if you use Lumenzia, you don’t have to think about any of this. Starting with v8.1, Lumenzia will automatically optimize your working gray space automatically for every document. Even if you are constantly switching between documents with different RGB profiles, Lumenzia will update the working gray space to the optimal result anytime you create a mask or selection. It supports all major (sRGB, Adobe RGB, ProPhoto RGB) and secondary RGB working spaces (eciRGB v2, beta RGB, ACES CG, REC 2020, any many more). If you edit your master file in an unknown RGB space (such as device-specific profile for your printer paper profile), the gray working space default to the safer gamma 2.2. By default, this option is enabled – so if you’ve updated to Lumenzia v8.1, you’re already getting the best quality results automatically on all your images.

If you wish to retain full manual control, Lumenzia also offers you that option too. Simply go to the Utilities menu in Lumenzia and disable the option to automatically optimize the gray working space and it will leave it alone.

Please note that Lumenzia never changes the RGB profile of your document, this feature only affects the working gray space to give you the best possible luminosity masks.

 

What to do if you use my free panel or another method for creating luminosity masks/selections?

Any time you work with an image which uses a different RGB color profile (the embedded profile is what matters, not the working RGB space), you should ideally update your Gray space to get optimal results. But updating it for every document manually is painful and prone to making mistakes. Instead, I would recommend one of the following approaches as the most reasonable solution for manually managing the gray space with minimal masking issues:

  • If you use ProPhoto RGB exclusively for all your images, set your working gray space to Gray Gamma 1.8.
  • Otherwise, set your working gray space to Gray Gamma 2.2 (even if you occasionally use ProPhoto, as the errors with gamma 2.2 are more tolerable as described above). Gray Gamma 2.2 is the right choice for Adobe RGB and is very close to the slightly more accurate sGray that should ideally be used with sRGB (note that you don’t have an sGray should for Photoshop CS6 or older, so Gray Gamma 2.2 is your best choice anyway). This is a very reasonable approach (prior to Lumenzia v8.1, this was my general recommendation, as manually updating for every document is tedious and prone to errors.)

These won’t give you perfect masks every time, but should be close enough to avoid the worse issues with mismatched gammas.

 

 

 

Black Friday sale

Black Friday is here! Now through Monday (Dec 2), you can save at least 25% off any of my products with discount code BF2019. This includes the Lumenzia luminosity masking panel for Photoshop, the Exposure Blending Master Course, and the new Dodging & Burning Master Course. You’ll save even more when you choose bundle upgrade options at checkout (see here for alternative bundles if you need help finding the specific mix of products you need).

Unlock the bonus course if you buy all three products by Dec 2

I have a secret exposure blending tutorial that I make available on rare occasions. This course is never for sale, you can only earn it. And you’ll earn it if you have purchased all three of my products (Lumenzia and both courses linked above) by the end of Monday, Dec 2. Any previous purchases you have made count toward this offer. But don’t delay, purchases made on Dec 3rd or later will NOT be eligible and there will not be another opportunity to get into this course for quite a while. No exceptions.

This tutorial includes a video tutorial on how to process the image shown below, PDF notes for the video, and the source RAW image – all hosted on the Teachable platform.

If you qualify, you will be notified via email when the course is added to your account next week. Please do not contact me about the bonus course before Dec 5, as I will be spending some much needed time with my family for the Thanksgiving holiday weekend here in the US and then will set up course access next week. If you have still not received notice by Dec 5, please contact me at that time for assistance.

What’s new for photographers in Photoshop CC 2020?

Adobe has just released Photoshop 2020 at Adobe MAX. As usual, there is a long list of updates to cover a variety of audiences. First, a quick note for those of you using my Lumenzia luminosity masking panel for Photoshop, it is fully compatible with Photoshop 2020.

 

Here’s a quick summary of changes that should be of greater interest to photographers (starting with what I consider the most impactful changes):

  • Auto-sampling with the New Content-Aware Fill. When you set the sampling area to “auto”, the green areas will be set more intelligently for you (instead of a generic green rectangle).
    • In my experience, this new auto option tends to do a very good job when you start with a reasonably isolation selection to target the area for fill. It’s a very nice enhancement that makes this powerful tool even faster to use.
    • See my previous Content-Aware Fill tutorial for how to make the most of this amazing functionality.
  • Transform Warp. When you warp (Edit > Transform > Warp or Cmd-T and click Warp Button), you can create custom grids to control transform much more precisely.
    • This offers some very nice functionality similar to what I demonstrated in my Perspective Warp tutorial.
    • Use the split or grid options in the toolbar to define your grid, and then drag intersection points (or their handles to rotate content).
    • You can select and move several points at once by <shift>-clicking and dragging a box around them with your mouse (or just <shift>-clicking multiple points.)
  • Erase while using Brush Tool by holding the “grave accent” key (this is the <`> key on the top-left of your keyboard).
    • While this may sound trivial at first (especially if you are using to using the <E> key to activate the eraser), this is actually very powerful because it gives you an eraser with the exact same settings as your brush (size, hardness, flow, opacity, etc). This makes it much easier to erase without constantly toggling settings to match what you just painted, especially brush size.
    • Note that when working on a layer mask, the eraser works in a very strange way. You would assume that it makes the layer mask black so that the layer becomes more transparent. However, the eraser actually works different on the layer mask than it does on the layer pixels. On the mask, it simply paints with the background color. If that happens to be black, you’ll get the expected results. But if it is some other color, that’s what will be used. So this may make your pixels less visible or more visible. If you see strange results, just click <D> while the layer mask is selected and paint colors will revert to black background (and white foreground).
  • Smart Object Selection Tool. This tool intelligently refines marquee or lasso selections around objects (ie, this is like having a “magic lasso” or “magic marquee” tool).
    • Do not expect a perfect selection, this is a tool to help you get to a good result more quickly. If you draw a reasonably well defined selection around a clearly defined object, it does a pretty good job of enhancing the selection for you to save time. You can then refine the selection as need (such as via additional selections or refine edge) to perfect the selection.
    • Turn on “Object Subtract” in the toolbar to use the same smart enhancement when using <alt/option> to remove part of the selection, such as removing portions of the interior of your initial selection.
  • Enhanced Properties panel now includes more capability to edit your layers (including alignment, rotation, flipping, remove background, and select subject) or text (numerous new properties).
  • Cloud documents. This allows you to edit the same document more easily across multiple devices (such as saving your work from your desktop computer and then opening it on your iPad), and is even designed to support offline workflows (so your changes are synchronized once you reconnect to the internet).
    • If you are going to use the new Photoshop iPad app, you should get familiar with this functionality to help more easily work across devices.
    • To save: Use the “Save as Adobe Cloud Document” option via File / Save As. Once you save a given document in the cloud, it will default to saving in the cloud going forward (unless you do another “save as” and choose different options).
    • To open: Use the “Open Adobe Cloud Document” option via File / Open. When you open a cloud document, any changes will go back to the cloud by default (unless you use “save as” and choose different options).
    • Cloud documents include support for offline workflows, meaning that you can open, edit, and save a document when you are offline. You may open a “cloud document” offline if it was recently used and your computer still has a local copy.
    • You may view your documents online in Adobe Assets.
    • Learn more on adobe.com.
  • Other changes:
    • Free transform lock is now sticky. The <shift> key is now used to toggle between constrained and unconstrained based on the lock status.This makes it easy to set a default, but you can always quickly click <shift> to change the behavior
      • Various adjustments (paths, masks, etc) are now treated in the same manner for consistency.
      • However, the crop tool does not use the same logic (it will follow any constraints you type into the toolbar, otherwise it is unconstrained by default and constrained if you continuously hold <shift>).
      • If you prefer the legacy behavior (ie unconstrained by default and constrained if you continuously hold <shift>), there is now a checkbox to enable “Legacy Transform” behavior under Photoshop’s General Preferences.
    • Improved Lens Blur.  Go to Filter / Blur / Lens Blur.
      • This filter now uses the GPU for much faster performance.
      • New click to focus (crosshairs icon) for images with a “depth map”. This gives you creative control over depth of field, such as found on many new smart phones. Depth maps are primarily saved with images from a Smart Phone with 2 or more cameras and appropriate settings to save an HEIC file (not something you’re going to get from your DSLR, though there are ways you can manually create one).
    • 32-bit Curves and Levels. If you are using 32-bit files (such as HDRs created inside Photoshop, not Lightroom), this is a welcome enhancement to use some critical tools without converting to 16-bit.
    • Faster new document interface.
    • Extract layers from a Smart Object. Just right-click the Smart Object and choose “Convert to Layers”.
      • Lumenzia has actually had this capability for a while (see this tutorial), but it is nice to see it in natively supported in Photoshop as well. Note, however, that the new Photoshop utility will not extract any saved paths or channels. So if you have been putting this content into the Smart Object (which Lumenzia can do to reduce the size of the parent document), you should continue using Lumenzia to extract preserve channels and paths from the Smart Object.
    • Select Subject” has been updated for faster results.
    • Open images directly from your iPhone (Mac only, sorry Windows users). Just go to File / Import from iPhone or iPad.
    • “CompCore” (which was introduced in PS 2019) has been further updated. This should be a seamless change, but there were some bugs when first launched in 2019. In the unlikely event that you see any issues with blend modes or other layer-related issues, try disabling CompCore via Preferences / Performance / Legacy Compositing.

 


See the store page for Lumenzia and course info.
See licensing for Commercial and Creative Commons (Non-Commercial, Attribution) Licensing terms.
Join my affiliate program.
See my ethics and privacy statement.