Bulk Background Removal: Process Thousands of Product Photos

You’ve got 2,000 product photos from last week’s shoot. Every single one needs the background removed before they can go on your website. Doing them one by one in Photoshop? That’s roughly 3-5 minutes per image if you’re fast. So about 100-170 hours of work. Over four straight weeks of someone’s time.

There are better ways. This tutorial walks through four approaches to bulk background removal, from free to enterprise, with real costs and quality comparisons.

The Four Approaches

1. Photoshop Actions + Batch Processing (Semi-Automated)

2. remove.bg API (Per-Image Pricing)

3. Photoroom Batch (Subscription)

4. Sirv.studio Workflow Studio (Pipeline + CDN Delivery)

Each has different tradeoffs on cost, quality, and speed. Let’s go through them.

Method 1: Photoshop Actions

Photoshop’s “Remove Background” feature (under Quick Actions) uses Adobe Sensei AI. It works reasonably well for clean studio shots. You can automate it with Actions and Batch processing.

Steps:

1. Open one image in Photoshop
2. Go to Window > Actions and create a new Action (click the + icon)
3. Name it “Remove BG and Save”
4. With recording active, do these steps:
   • Properties panel > Quick Actions > Remove Background
   • File > Export > Export As (PNG, your desired settings)
   • File > Close (don’t save)
5. Stop recording
6. Go to File > Automate > Batch
7. Select your Action, set Source folder (your raw photos) and Destination folder
8. Click OK and let it run

Cost:

$22.99/month for Photoshop (you probably already have it)

Speed:

About 10-15 seconds per image on a decent machine. 2,000 images = 6-8 hours unattended.

Quality:

Adobe Sensei does a good job on clean studio shots, but it’s inconsistent with complex edges. Hair, fur, transparent objects, and thin straps often need manual cleanup. For a batch of 2,000, expect to manually fix 10-20% of the results.

The catch:

Your computer is tied up for hours. Photoshop can crash on very large batches. And there’s no way to chain additional operations (like resizing for different marketplaces) without creating separate actions.

Method 2: remove.bg API

remove.bg was one of the first AI background removal tools and it’s still one of the best for pure quality. Their API lets you process images programmatically.

Steps:

1. Sign up at remove.bg and get an API key
2. Install their CLI tool: pip install removebg
3. Put all your images in a folder
4. Run:

for file in ./input/*.jpg; do
  curl -s -X POST "https://api.remove.bg/v1.0/removebg" \
    -H "X-Api-Key: YOUR_API_KEY" \
    -F "image_file=@$file" \
    -F "size=full" \
    -o "./output/$(basename $file .jpg).png"
  sleep 1  # rate limiting
done

Or use their Python SDK for parallel processing:

from removebg import RemoveBg
import os

rmbg = RemoveBg("YOUR_API_KEY", "error.log")
input_dir = "./input"
output_dir = "./output"

for filename in os.listdir(input_dir):
    if filename.endswith(('.jpg', '.png')):
        input_path = os.path.join(input_dir, filename)
        rmbg.remove_background_from_img_file(
            input_path,
            new_file_name=os.path.join(output_dir, filename)
        )

Cost:

$1.99 per image (full resolution) or $89/month for 500 credits. 2,000 images at full res = $3,980. With the subscription: $89 x 4 months = $356 (500 credits/month).

Speed:

2-5 seconds per image via API. 2,000 images = about 2-3 hours with rate limiting.

Quality:

Best-in-class for single-image accuracy. Handles fine hair, semi-transparent objects, and complex edges better than most competitors. Manual fixes needed on maybe 5-10% of images.

The catch:

Expensive at scale. $3,980 for one batch is hard to justify unless you’re a large retailer. The subscription model is better but you’re limited to 500 images/month on the base plan.

Method 3: Photoroom Batch

Photoroom offers batch processing through their web app and API. The quality is close to remove.bg for most product types.

Steps:

1. Go to photoroom.com and sign up for Pro ($9.99/month)
2. Click Batch Editor in the dashboard
3. Drag and drop up to 100 images at once
4. Select “Remove Background” as the action
5. Choose output format (PNG for transparent, JPG with white bg)
6. Download the results as a ZIP

For larger batches, repeat the process or use their API.

Cost:

$9.99/month for Pro (unlimited background removal via web app). API pricing is separate and per-image.

Speed:

3-8 seconds per image through the web app. Uploads are the bottleneck for large batches.

Quality:

Good. About on par with Photoshop’s Quick Action, slightly below remove.bg for tricky edges. Excellent for clean studio shots on white or light backgrounds.

The catch:

The web interface maxes out at 100 images per batch. For 2,000 images, that’s 20 batches of uploading and downloading. The API removes this limit but costs extra. No way to chain additional operations (resize, optimize, deliver) in the same workflow.

Method 4: Sirv.studio Workflow Studio

Sirv.studio takes a different approach. Instead of just removing backgrounds, you build a pipeline that chains multiple AI operations together and runs them against entire folders.

Steps:

1. Upload your product images to your Sirv account (drag and drop, FTP, or S3 API)
2. Open Sirv.studio and go to Workflow Studio
3. Create a new workflow:
   • Step 1: Background Removal
   • Step 2: Auto-crop (trim empty space around the product)
   • Step 3: Resize to 2000x2000px (or your target spec)
   • Step 4: Optimize for web delivery
4. Select the folder containing your images
5. Run the workflow

The processed images land in a designated output folder on Sirv’s CDN. They’re immediately available via URL with automatic WebP/AVIF conversion.

Cost:

10 free credits/month. $9/month for 100 credits. 1 credit = 1 AI operation. A 4-step workflow on 2,000 images = 8,000 credits. At bulk pricing, roughly $400-$600.

Speed:

Processing happens on Sirv’s servers. A batch of 2,000 typically completes in 2-4 hours depending on image size and queue.

Quality:

Background removal quality is comparable to Photoroom. The real advantage is the pipeline: background removal + cropping + resizing + optimization in one pass, with output going directly to CDN.

The catch:

Credit-based pricing means you need to plan your budget. The background removal alone isn’t better than remove.bg. The value is in the pipeline and CDN integration, not the individual AI quality.

Cost Comparison

Here’s what each approach costs for a 2,000-image batch (background removal only):

Background removal only. Photoshop and Photoroom assume existing subscriptions. Sirv.studio estimate for BG removal credits at volume pricing.

Tool	Cost (2,000 images)	Time	Quality (1-10)	Pipeline?
Photoshop Batch	$23/mo (existing sub)	6-8h (your machine)	7	Manual actions only
Photoroom Pro	$10/mo	4-6h (20 batches)	7.5	No
Sirv.studio	~$180 (credits)	2-4h (server-side)	7.5	Yes (multi-step)
remove.bg API	$3,980 (full res)	2-3h	9	No

Which Method Should You Use?

Tight budget, small batches (under 100 images): Photoshop. You already have it, and the quality is fine for clean studio shots. Manual fixes take time but there’s no per-image cost.

Regular batches, subscription budget: Photoroom Pro at $10/month is the best value for unlimited background removal. The web interface is clunky for large batches, but for regular shoots of 50-200 images, it’s hard to beat.

Large catalogs needing full pipeline: Sirv.studio Workflow Studio. If you need background removal plus resizing, cropping, and format optimization all at once, with output going directly to a CDN, this is the most efficient option. The credit cost is higher than Photoroom for background removal alone, but you save time on all the downstream steps.

Quality is everything: remove.bg. For product categories where edge quality really matters (jewelry, hair accessories, transparent products), remove.bg’s results need less manual cleanup. Expensive, but the time savings on manual fixes can justify the cost.

Tips for Better Results (Any Tool)

1. Shoot on a contrasting background. White product on white background = poor results everywhere. Use a light gray or blue backdrop if your product is white.

2. Consistent lighting kills shadows. AI tools handle product edges well but often leave shadow artifacts. Diffused, even lighting from a lightbox minimizes this.

3. Check edges at 200% zoom. AI background removal looks fine at normal size but reveals issues when zoomed in. Spot-check at least 10% of your batch.

4. Keep source files. Always keep your original photos. AI tools improve over time, and you might want to re-process older images with better algorithms later.

5. Test with 20 images first. Before running 2,000 images through any tool, test with a representative sample of 20 (including your trickiest products). This saves you from discovering quality issues after processing the entire batch.

6. White isn’t always white. “White background” for Amazon means RGB 255,255,255 within 5% tolerance. Some AI tools leave a very slight gray cast that fails Amazon’s automated checks. Test your output against marketplace requirements specifically.

Your Next Step

Pick your trickiest 10 product photos (transparent bottles, fine jewelry, white products on light backgrounds) and run them through two different tools. Compare the edges at 200% zoom. That tells you more about real-world quality than any marketing page.

For most e-commerce teams, the workflow is: Photoroom or Sirv.studio for daily batch processing, with remove.bg as a fallback for the 5-10% of images that need top-tier edge quality.